development

[frosty hemlock]

My first every PR to an opensource project

[wide reef]

Hoped you had fun!

[frosty hemlock]

Definitly did, I will see if I can do more of these "side quests"

[wide reef]

You're welcome to venture into formatting and checking the JS code for data handling inside the HTML pages. I tend to use a lot of AI when I build front-ends and give it minimal attention when possible

I generally enjoy the design part of things and can do good backend services, but frontend is beyond me

[frosty hemlock]

I generate frontend too lmao

I have also never really used JavaScript, I mainly do TypeScript

And even more of Java (Spring Boot)

[wide reef]

Whatever makes this thing better front-end wise

you're welcome to try

[frosty hemlock]

I will keep an eye on this project

[wide reef]

I'll mostly focus on stuff I'd like to see generally. So library and blog management mainly

[wide reef]

Books are now generally available on dev!

Although you'd have to build the container yourself

[frosty hemlock]

@wide reef Does dev work on your side?

[wide reef]

It does, what issue you're facing?

[frosty hemlock]

I click on a blog and its empty

[wide reef]

Screens?

[wide reef]

Oh yeah, one of the color validation things had missing ident

give me a sec

Fix is pushed

[frosty hemlock]

Yup working now

testElem.style.color = testColor;
// The browser sets style.color to an empty string if invalid
if (testElem.style.color !== "") {
    element.style.color = testColor;
}

I didn't know about this one, thats interesting

[wide reef]

It's likely one of the automated triages. What's on your mind?

[frosty hemlock]

I thought invalid colors would just give an exception

Never thought about it too tbf

[wide reef]

I will ignore the color from what I am reading

[frosty hemlock]

If the color is an empty string it will make a console.warn()

[wide reef]

but warns are not regarded

so there is essentially no error happening and it just dies lol

[frosty hemlock]

Yeah lol

[cedar wolf]

can someone from LTT Staff send me a 3d render of the best motherboard cpu gpu and ram that you guys have?

[hollow basalt]

Lmao

[grim frost]

Subject: Feedback Request – Thinrex OS Development

Hey everyone,
I’m currently developing Thinrex OS, a next-generation security-focused operating system designed for privacy, performance, and universal compatibility. It’s built on a minimal, hardened Alpine-based foundation with Wayland (Sway) as the default desktop and features AppArmor, Firejail sandboxing, and full virtualization support for isolated application execution.

Thinrex also integrates modern app ecosystems like Flatpak, Wine, and Waydroid, ensuring both native and cross-platform software support. The goal is to create a secure-by-default desktop OS that feels fast, futuristic, and safe for everyday users, developers, and professionals.

I’d appreciate technical and architectural feedback on:
• System hardening or sandboxing improvements
• Better virtualization isolation models
• Package layout or build structure
• GUI installer (Calamares) customization ideas

Any thoughts or suggestions to refine Thinrex are welcome.
Thanks in advance for your insights.

— Fell Flow (Thinrex Dev)

[bold prawn]

whats your target audience for that really / what problem are you trying to solve that other distros dont already do? (assuming this is not just a fun project / an exercise for you)
everyday users wont play well with a tiling wm. advanced users might prefer the flexibility of arch or nixos to customise it to their needs. it seems like you dont really know exactly where this is supposed to fit in. if youre just doing it because you want to, thats completely fine. if you do want an audience for it, you need to think who that is and what their needs are, to create a sensible package for them.

[grim frost]

The target audience is people who care about their security and privacy but don’t wanna give up their apps they use on a daily basis one thing that my OS has that others don’t is that every app is like its own VM. It cannot escape it and it’s also sandbox.

[bold prawn]

have you checked out nixos?

all apps are declarative builds with their own environments

[old arrow]

qubes os
silverblue
arch but you install flatpak and qemu

[grim frost]

I have not does it have built in tor over VPN with dns protection

[bold prawn]

nothing is really built in, you can customise it to your needs and simply install a nix package for that

[bold prawn]

also how do you achieve the isolation of every app being its own vm, do you run them in docker containers? qemu? lxc?

[grim frost]

I think it’s firecracker one sec

[grim frost]

Thinrex runs each app in a KVM micro-VM (QEMU). Not Docker/LXC.

How Thinrex isolates apps
• Process boundary: one QEMU per app.
• Kernel boundary: each app gets its own Linux kernel.
• RootFS: immutable base image + per-app writable overlay.
• Host↔guest IPC: VSOCK.
• Files in: virtio-fs read-only mounts for app bundles.
• Network: off by default. Opt-in per app.
• Display: forwarded from guest to host (waypipe/xpra/spice), GPU off by default.
• Surface minimization: -nodefaults, virtio only, -M microvm when possible

[bold prawn]

how do you handle resource management with that? id imagine with everything running in qemu there is a huge performance penalty and youd need substantial overprovisioning on each VM. each app running its own kernel probably pushes up the memory consumption considerably, seeing how typically youd run dozens of apps. which is why docker does not do that by default.

[old arrow]

holy chatgpt

[grim frost]

You’re right: naïve “one VM per app” wastes RAM. Thinrex stays efficient with four levers: trim, share, cap, and reclaim.

What keeps overhead low
• Trim: tiny kernel+initrd, -M microvm or q35 -nodefaults, virtio-only, no BIOS devices. Idle VM <120-180 MB is realistic.
• Share: read-only app bundles via virtiofs. With DAX in guest, file pages are mapped from the host page cache, so multiple VMs share them.
• Cap: start each QEMU in its own cgroup with CPU and memory limits.
• Reclaim: virtio-balloon for overcommit and quick give-back; host KSM to dedupe anonymous pages; zram to compress cold anon.

[bold prawn]

hm, ok. but you said you use a microvm with firecracker. and then you said you use qemu. which one is it?

[old arrow]

hint: the cap part is meaningless when a virtualizer like qemu inherently sets a static memory limit of the vm

[eternal ferry]

there's a memory deduplication feature you can enable, its fairly effective

it's still wasteful compared to containers but it does help quite a bit

[bold prawn]

i feel like before we get into these details, the basic info already doesnt quite add up

[old arrow]

that's fair

[bold prawn]

youre still absolutely right of course :)

[old arrow]

yeah but containers are scary 👻 no wall of fire between your kernel and the arguments of fopen()

[eternal ferry]

well it is october

[bold prawn]

id say with the whole world running on docker and kubernetes now, we can take a moderate amount of spookiness at this point

[eternal ferry]

i run everything in its own VM usually

[old arrow]

somebody will find a vulnerability and steal all of google's managed kubernetes engine computers

[grim frost]

Ya I said I think it uses firecracker

[old arrow]

the only firecracker i know is https://milktooth.com/products/firecracker

[grim frost]

This is what it says You’re using QEMU (TCG) to boot a GRUB2 hybrid BIOS/UEFI ISO that starts an Ubuntu 6.8 kernel with a casper initrd and mounts a SquashFS live root. No Firecracker. No Docker/LXC.

Stack, layer by layer
• Hypervisor: qemu-system-x86_64 with -accel tcg and -cdrom Thinrex-boot-*.iso.
• Firmware: BIOS by default; optional OVMF for UEFI.
• Bootloader on ISO: GRUB2 (/boot/grub/grub.cfg).
• Kernel: /boot/vmlinuz from the chroot (6.8.0-31-generic in your logs).
• Initrd: /boot/initrd.img built by initramfs-tools + casper.
• Root FS: /casper/filesystem.squashfs (read-only live root).
• Userspace: minimal Ubuntu base; single VM environment.

[bold prawn]

yes, but you also said it uses qemu. which to my knowledge you cant just combine. its either your apps use firecracker OR qemu.

ok, so when you said you think it uses firecracker that was simply a mistake then

how do you deal with apps that need hardware acceleration and access to my storage. lets say i want to start blender and edit a file on my computer. do i have to manually create a mount and change the settings of my blender app to enable gpu passthrough?

[grim frost]

It uses QEMU only right now. Firecracker is not in the path. A single VM can’t “mix” QEMU and Firecracker; each app instance runs under one hypervisor at a time. A platform can support both as switchable backends, but not simultaneously for the same VM.

[bold prawn]

i know, thats why i was confused when you said firecracker in one message and qemu in the next

[grim frost]

I just figured that out here’s How Thinrex handles GPU and files
• Per-app VM. Each app runs in its own minimal VM. Dev uses QEMU. Prod targets Firecracker. Same image, different launcher.
• Files auto-mounted. The app manifest declares host folders. The runtime shares them via virtiofs. Inside the VM they show up at /mnt/host. Read-only by default, can be rw if the app asks. No user steps.
• GPU accel defaults.
• Default: virtio-gpu + virgl → hardware-accelerated OpenGL for most GUI apps.
• Need CUDA/ROCm or max perf: VFIO PCI passthrough to a real GPU. Optional and opt-in.
• Fallback: software rendering if neither is available.
• UX example (Blender). Launches in its VM. Your project folder is already at /mnt/host. Open/save works on host files without reconfig.
• Resource control. Per-VM caps for vCPU/RAM/IO. virtio-balloon for reclaim. cgroups for host fairness. Images are slim to keep per-app RAM low.
• Security. No devices or folders exposed unless whitelisted in the manifest. Network is per-VM and policy-controlled.

Bottom line: You don’t mount anything or toggle GPU by hand. Apps get isolated VMs with your files auto-available and GPU accel on by default. Passthrough is there when you need heavy GPU

[old arrow]

having gpu acceleration in vms is very optimistic, it barely works if at all

[grim frost]

Yes I’m very optimistic I like to try to break boundaries it’s who I am

[old arrow]

i like the spirit but it doesn't really help when it just doesn't really exist

i think the current state of the virtio drivers is that it should boot and work for some 3d applications sometimes

[bold prawn]

yes, if you want to give users more safety in their everyday lives, their everyday apps still need to work and without major issues or major performance drops. if youre safe but your apps arent usable, theres no point.
i would say: identify common use-cases, common apps and do performance tests as well as actually using them to see if they work just fine and perform decently. a performance penalty for more security is often perfectly acceptable, as long as its a reasonable penalty.

[weak pilot]

why does an “os dev” not know how to use screenshot software?

[grim frost]

That computer is not with me and I’m developing a os with note I’m not that good at coding I use codex and regular ChatGPT

[weak pilot]

you should start with any other project then

maintaining an operating system isnt a starter one

[grim frost]

I know but I’m going to keep working on my project

[grim frost]

I’m confident

[grim frost]

I have to develop my own driver

[weak pilot]

also you list cuda on a linux distro that uses musl?

are you even aware of that incompatibility

[grim frost]

CUDA is dynamically linked against glibc and requires the full GNU C library ABI for its runtime and driver APIs.
musl libc — used by Alpine and Thinrex — does not implement some glibc-specific syscalls and symbols that NVIDIA binaries depend on (pthread_cancel, __libc_start_main, malloc_trim, etc.).

Thinrex deals with this incompatibility like this:
1. Split ABI zones:
• The base system (init, sandbox daemon, shell, core tools) runs on musl for compactness and security.
• GPU-dependent workloads spawn inside a glibc-based microVM (Ubuntu or Debian minimal image).
• This lets you use CUDA, ROCm, or OpenCL stacks without polluting the musl root.
2. Runtime broker:
• The sandbox daemon detects GPU-capable binaries through ELF inspection.
• If CUDA or OpenCL symbols are present, it launches the app in a glibc guest microVM (Firecracker or QEMU-lite).
• Host /dev/nvidia* and /dev/dri/* are passed through with seccomp rules.
3. Filesystem mapping:
• The sandbox manager mounts host project directories as 9p or virtio-fs shares so GPU workloads can read/write user files without leaving the isolation boundary.
4. Performance:
• GPU-bound tasks aren’t slowed by this indirection because CUDA runs directly on the host kernel driver; the microVM only virtualizes the userspace runtime.
• CPU-heavy code is still compiled with musl and runs natively for better density.

So CUDA isn’t linked into musl directly; it’s offloaded to a glibc microVM runtime for compatibility.

I just wanted to say thanks for the support I truly appreciate everyone of you

[undone arrow]

what's a standard place for programs to place downloaded files? like a game launcher for example

[austere bobcat]

%USERPROFILE%/Downloads? Or not that kind of downloads

[shrewd valve]

almost completely up to them, basically anywhere where the user has access to without admin

[undone arrow]

i see. Im just thinking about where i would put my default location as for my game launchers files

[keen sorrel]

Windows convention for system-wide would be under %ProgramData% while user specific would be under %LocalAppData%. %TEMP% could also be viable depending on the context.

[undone arrow]

the context is game files, so I assume temp won't be good for that. perhaps programdata?

[keen sorrel]

yea. %ProgramFiles% is another popular choice, especially for long-lived files. The typical convention is for %ProgramFiles% to be generally read-only requiring admin to write, while %ProgramData% is read-write. Though really you're free to handle it however you'd like

[undone arrow]

awesome. thank you! will I need to request elevated privileges then to download into program files?

also, I assume these constants aren't multiplatform

so they wont work on Linux for example

[keen sorrel]

You can adjust the folder permissions to whatever you'd like during initial install. The installer would need to be already elevated to install into %ProgramFiles%. The only thing "enforcing" this convention is the default inherited folder permissions.
Linux has its own conventions for where to place things. ~/.local, ~/.cache, or another custom sub-dir of the user's home ~ for user specific files. An appropriate subdir of /var is the usual place for system-wide stuff
https://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/var.html

[undone arrow]

cool, thanks so much for the info 😃

[marsh oasis]

Hey, not sure if this is the best place to ask... I'm having some problems with having my website hosted and it's quite bizarre imo. Lmk if y'all have any solutions or know a better place I can ask this.

It's an Oracle cloud server that is hosting a Node.js Express server on Ubuntu. After many troubles (very dumb problems with Oracle's firewall), I was finally able to get to host the website on the server's public IP and that works amazingly.

I've shifted the DNS to cloudfare and I created an A record to the server's public IP. Everything, is theoretically set up perfectly but, I can't access the site from the domain...

Happy to provide any other information required :)

[vapid grove]

Have you done a route trace to see where it's failing?

[mental panther]

Out of all the many providers you could choose... Why oracle cloud?

[mental panther]

so it's still accessible from the IP but doesn't work from the domain? What error are you getting?

is your computer/network's DNS set to CloudFlare? If not you'll have to wait longer for your records to update

[marsh oasis]

Lmao. It's free

[mental panther]

Aws has a free tier too iirc... And GitHub pages sounds like it'd work aswell and that's free last I checked

Anyway back to the actual topic lol

[marsh oasis]

My computer is using cloudflares DNS actually, it did seem to be a DNS caching problem cause many other people checked it out, not many people from my country were able to access it though

[marsh oasis]

AWS free tier is only for a year, github pages is just static

[mental panther]

Wait is my life a lie? I swore GitHub pages could run a node app

[marsh oasis]

For real? Oh well, isn't it like really low RAM? I like to cache my DB in RAM hahahaha

[mental panther]

from a quick google my life is indeed a lie

Unless there's some other non-pages thing that does it, idk

[marsh oasis]

I mean it's kinda in the name, no? Pages -> hosts some pages nothing more nothing less

[neon oriole]

Okay its not free hosting but it really is the next best thing: Heliohost has a plan that allows for a 1 time payment(a donation) , it comes with plesk and softaclous, and all the usual 🙂

[marsh oasis]

I have no idea what actually was the solution but I set up SSL certs properly, turned it on and off a couple of times, fucked around with some cloudflare settings, and let DNS caches die out... mostly working all fine now!

I don't think it has anything to do with the host anyways

[rancid crane]

hey guys i need your help on my app

[austere bobcat]

https://dontasktoask.com

[rancid crane]

im trynna create my app but im stuck

[austere bobcat]

Stuck on what

What kind of app?

[rancid crane]

the idea behind this app is to make jewellery shopping simple, enjoyable, and accessible to everyone. Instead of walking from shop to shop, people can browse beautiful accessories right from their phone view clear pictures, compare prices, and choose what matches their style.

Once they find what they love, they can easily add it to their cart and pay securely through M-Pesa. The goal is to create a smooth, modern shopping experience that feels personal where elegance meets convenience, and every order feels special.

M-Pesa is a mobile money service that lets people send, receive, and store money using their phone no bank account needed. It’s widely used in Kenya and several other countries.

I was creating a jewellery ordering website where customers can browse different pieces, add them to a cart, and check out using M-Pesa for payments. I built it using Visual Studio Code (VS Code) that’s where I was writing and editing my HTML, CSS, and JavaScript files.

so i had two >script> sections with duplicate logic and i got stuck i didnt know how to remove it

[old arrow]

Probably netlify?

[frosty hemlock]

@cloud knotI have tested the 9060xt 16GB for AI and I must say, im not dissapointed

Its not as fast and as well supported like Nvidia but it still does a pretty good job at just general AI models

Even some image generation works surprisingly well

[frosty hemlock]

Stable diffusion 3.5 medium

[soft fern]

flux models are really good for image generation too btw, I kind of stopped messing with stable diffusion after the XL/refiner stuff came out, could get lots of detail with various Loras from civitai and whatnot but flux just generally does a better job out the box from what I've seen, ComfyUI good for messing around with whatever though the templates make it pretty easy to load things up and try them out

Occaionally still get the "ouroborus" shrimp but overall it does a pretty impressive job especially when comes to things like diffraction/reflections, it isn't always right perse but interesting to see what it comes up with at times... unfortunately I can't eat this picture and the embodied AI can't cook it up for me yet

[cloud knot]

i still haven't found time to figure out AI image generation stuff yet.

[soft fern]

comfyui makes it super easy is like ollama but for image gen models, handles most of the setup for dependencies and they have a bunch of prebaked "templates" for node setups that load models and run prompts through them, I just loaded up flux dev here to poke at and been using wan2.2 a little bit for messing with vid generation too, don't really have a use case at all for the latter yet but just interesting to see how things are progressing and what you can run local

they still will fail at stuff like this, I asked for "a wine glass already completely filled, being filled over the brim from a bottle" it always shows wine glasses half full basically because so many of them look that way when labelled in the dataset, overflowing wine glasses aren't common enough for it to adhere to the prompt

I think the vid gen stuff in it's current form probably good for like story board or "previs" type stuff to get some idea of how a scene would look or whatever but seems like a lot of patchwork to use it more directly still

[cloud knot]

i remember how in early days of bing image generator was taking image prompts literally instead of intepreting it as food names, so "rooster with wine" ended up with an actual rooster standing on top of wine bottle 😄

[soft fern]

hehe yeah can still have misunderstanding of meaning as well but a lot of the default models feel more "grounded" in reality to some degree at least

still end up with things like anime characters with some random ultra photoreal body parts of other kinds of weird blending of styles etc. but can be fun, think stuff like words and symmetry have improved but still not perfect

the vid generation things similarly are getting more impressive in terms of simulating real physics but still lots of limitations in terms of coherence over time other issues, is similar in terms of the success to fail with running LLMs and image gen, sometimes impressively crazy sometimes falls on it's face

this is "will smith eating plate of spaghetti" from wan 2.2, it apparently didn't know the will smith I was talking about 🙂

the spaghet sort of gets vaporized in his mouth but it's definitely closer than the early attempts

[wraith turtle]

Anyone have knowledge of compression and/or embedded programming mind sanity checking something I'm writing?

It's not actual code, more of a design spec/theory

[soft fern]

A little bit and a little bit but it depends on what kind of spec/theory it is

[wraith turtle]

Using a four layer pipeline for compression using a mix of deflate 9, exact deduplication and similar deduplication to see if we can push the current density and compression expectations on an extremely low budget (mcu based)

https://github.com/1Jamie/HMSE

[undone arrow]

im building myself a game launcher using avalonia, and I want to be able to have patch notes readable on one of the windows. my question is how should I go about this? I was thinking just a text file on my server which I keep appending patches to with markdown but that feels a bit ugly and annoying. is there a better way to do this?

[old arrow]

That's probably the most common way. And you put newer updates towards the top, so when it opens people see the newest changes first

[cloud knot]

static file generator from the .md file and then show HTML on your client ?

In general, patch notes are really just text files which you manually edit 🤷‍♂️

[cosmic hatch]

Am I the only person writing pages and blogs and more in plain HTML+CSS

[fading osprey]

I generally use markdown for stuff because it's easy

[weak glen]

My website is more or less plain HTML+CSS, but that's mostly just because it's how I already know to do it, and it doesn't really matter since it's a static website anyways

[old arrow]

Templating and being able to put global components on the entire site is pretty important to me so I can't just write a bunch of manual HTML files in my public dir

I use Hugo

[uneven swallow]

probably im gonna learn C++

[covert axle]

yay

c++ is awesome

[cosmic hatch]

learn rust

[hollow basalt]

scratch

[weak glen]

machine code

[uneven swallow]

https://tenor.com/view/monster-versus-vs-alien-monster-vs-alien-gif-16037253809360562033

[hidden patrol]

Convince me of why to learn rust. I’m looking for a new language to learn.

[old arrow]

what's your current programming experience?

[hidden patrol]

PHP, VB, C++, Java, Ruby, limited Python.

All self taught

[old arrow]

rust should be pretty interesting compared to the rest of those. The style is very modern and has newer patterns (errors as values, stronger memory safety structures are the things that come to my mind most)

[hidden patrol]

Where would you rate it in terms of complexity to learning compared to Java or c++?

[old arrow]

definitely closer to c++ than java

i feel like it's similar to c++ where it doesn't hide things so much like other higher level programming languages

this one is more as a joke

[hidden patrol]

I’ll give it a look this weekend

[old arrow]

i think the best thing is if you have the motivation to actually persevere when it gets hard, because i see lots of people that just quit out of rage and have really messed up conceptions of what the language is actually like

the same goes for C and C++

[uneven swallow]

https://tenor.com/view/shocked-monkey-stan-twt-macaque-monkey-reaction-my-honest-reaction-gif-4617235472778414495

[weak brook]

try FORTRAN

[fading osprey]

where my COBOL recommendations at? come on guys!

[uneven swallow]

isnt it old?

[weak brook]

yes.

but honestly, you'd be suprised how much random stuff still uses things like FORTRAN and COBOL

[uneven swallow]

true

[limpid reef]

Sure, but a large portion of Western Society's financial and insurance institutions rely on it... Which is becoming more and more of a problem as people fluent in FORTRAN / COBOL die.
https://www.theverge.com/2020/4/14/21219561/coronavirus-pandemic-unemployment-systems-cobol-legacy-software-infrastructure

[uneven swallow]

bruh

[cloud knot]

but only because of financial institutions idea of "if core system ain't broken, don't touch it" mentality. Instead of dedicating a decade to mirror the system in a modern language they hope it will work out, that they will find replacement COBOL programmers. And when that won't happen, they will rush the replacement they could have planned for.

[crystal apex]

Cant really say any more than this but I have been told that this process has been happening for a while now at some larger institutions

[cloud knot]

banks are switching over in non-critical systems, but critical transaction systems are probably going to be problematic even for relatively modern banks

In 2024, the IRS announced a transition from COBOL to Java thanks to the Digital First Initiative.
When IRS can't find enough money to switch over, who can 😄

[weak brook]

also not really worth switching over to it if it'll cause more problems than good. older languages should still be used, even though they might not be as feature-rich

if it works, it works. This is why we still used FORTRAN for some radar applications.

you still see Ada used in flight control software.

[violet bay]

Ie the whole banking sector working on a archaic COBOL core with a bunch of different API’s bolted on to make it work with modern online banking etc

[cloud knot]

issue is, requirements change. And "if it works" is a problem when you want to make a change, but you got nobody to do it. Or at so expensive rate, it is not worth it anymore

[weak brook]

Is this a problem with the language, or is it a problem with schools not teaching it?

I used to have a FORTRAN elective. They stopped it because "it isnt useful."

[cloud knot]

nobody teaches COBOL for decades, it is an outdated language, but it is very good for math

[weak brook]

I remember we used to use FORTRAN for a lot of signal processing stuff in the radars i used to work on.

It was a mix of it being super efficient, and having a lage amount of proven code already existing.

C has caught up, though.

[random igloo]

Heyy! Chat

I know this may come off as advertising / spam but I do think figured I will shoot my shot ;D

Hey!

https://lovable-x-zerobase-hackathon.devpost.com/

(If you can't sign up, use this link: https://tally.so/r/n0X1z9)

Hosting a hackathon, 24 hours - 1K USD cash prize :))

It's happening simultaneously in Korea & Online!

November 9th 👀

P. S. It's free to sign up ;D

[grand shadow]

hi

[nocturne galleon]

I'm 38 and trying to get in at entry level, can't get even get an interview except with the public sector. I know Java and some C#, MVC patterns for both Spring Boot and EFCore. Am I cooked?

Would I be wasting my time to start learning React I wonder.

[vapid grove]

What would you be looking do? Having some sort of goal or focus will help build a path

Wouldn't say you are cooked, but its not going to be easy in the current times

[nocturne galleon]

I appreciate your reply, thanks! I don't really know, I learned through game dev, went to college a few years ago and stormed it. Doing a postgrad in fintech (a lot of python). I think I graduated at just the wrong time since AI just came out. Doesn't seem like basic MVC + crud is enough now, but I did actually get an offer after college, just couldn't meet the reference requirements. It's been utter shit since then, no interviews at all.

My college degree was pretty thorough, MQTT and gRPC, algos and datastructures, had it all, so I don't think I'm quite cooked.

[vapid grove]

I'm a game dev (and games adjacent) developer 👀

[nocturne galleon]

How are you finding it?

[vapid grove]

Being doing for 18 years so been through ups and downs

Been mostly pretty good tbh

Don't really have specific advice beyond the usual on ensuring that you've had your linkedin and resume reviewed, examples of work is always handy especially if it's novel

I don't know where/how you are applying but if you aren't already, I would use ATS site queries (like this one https://jobgen-waahjt.manus.space) where it searches directly on career sites of companies as well as looking off the beaten path a bit.

Startup job boards like

• https://jobs.a16z.com/jobs
• https://www.ycombinator.com/jobs

I do have more game related job advice compiled here too https://docs.google.com/document/d/1rDIupep9HG7u6rcsQBHnJ2v1J3TL9ZjpfRrM3OME9Xs/edit?tab=t.0#bookmark=id.6viav36a49o2

[nocturne galleon]

Bro, thank you for all that! Maybe I should look at game dev specifically. I don't care about crunch, and I have other skills since I used to be a composer/sound designer.

Is Unity still the go to or is it Unreal now that Unity had that massive stupidity with fees?

[vapid grove]

Both are still valid. Really generally: Indie and Mobile = Unity. AAA/AA = Unreal

[nocturne galleon]

@vapid grove Thanks for taking the time and showing me those resources, that doc is really interesting! I might have been sleeping on opportunities I didn't know existed.

[frosty hemlock]

Just learn C

And make everything in C

[old arrow]

unfortunately doesn't change the cold truth that operating system behavior, systems programming, and computer science are very complicated

[frosty hemlock]

True

I made my own programming language in C one time

[old arrow]

all options are terrible

[frosty hemlock]

I like C for how low the learning curve is, it took me about a week to really learn

But I love Java too, for how structured it is but I do think Java is a bit verbose

[nocturne galleon]

I feel like Python takes it too far in the other direction, not sure about C. Manual garbage handling... hrm...

[frozen flame]

[old arrow]

the loop returns to rust

[wraith turtle]

anyone want to try a dumb web based arpg i made today when i was bored?

im kinda curious on the feedback portion as i did some stuff i dont usually spend the time on

[limpid reef]

Sure. Reply with the GitHub link to your code and latest release.

[wraith turtle]

I guess I can throw it there lol I just have it in a zip 😂

[frozen flame]

Downloading a zip file from a stranger on the internet seems safe

[wraith turtle]

that's why i didn't argue with them

[wraith turtle]

https://github.com/1Jamie/shape_slayer doesnt have a release though as its not far enough to have one 😅 that and you just need to clone it and open the index.html

actually i will do one even better so no one can complain 😉

https://1jamie.github.io/shape_slayer/
this url is the actual directly playable game

[keen citrus]

its actually pretty cool

[fading osprey]

quite enjoyable actually, good work

[cosmic hatch]

hey that's actually reaally cool, well done

[wraith turtle]

Aww, ty ty @cosmic hatch @fading osprey @keen citrus

[keen citrus]

yeah seems fairly well polished, the first boss is very tough especially when you dont have alot of health

and it took me until the 3rd run to realise that the things at the bottom i could use with buttons, i just only figured out the whirlwind thingy

so maybe having some sort of indicator for those, and reducing the difficulty of the first boss might be good?

[wraith turtle]

Sounds good, thank you for the amazing feedback! I've been thinking of increasing the XP gained a bit so you level up more often a little bit to fix the health issue though yeah, honestly the first boss I was pretty proud of though .. probably shouldn't be the first boss lol

Though I feel like the next thing I plan to implement might fix all of the above

I want to initialize a spawn room and introduce rogue like in-between run level ups. So every run you earn a currency when you die you can use to permanently upgrade your speed, health or attack which honestly might solve both issues

That would be the limit to force you to die the first few times and level up

I want the bosses to be truly challenging without being unfair which is a hard balance (also I just haven't built the tutorial or tool tips for the class selection so that's kinda on me for not introducing players to the game properly)

[wraith turtle]

there was a big kid update today to the game XD
https://1jamie.github.io/shape_slayer
its got a lot of balacing changes, now has mobile controls (not the best but i need feedback on them) and graphics overhaul
also introduced the rougelike mechanics with the start room/nexus with currency earned each run you can apply to upgrades

[cosmic hatch]

add rounding when displaying Game.currentcurrency

[hollow basalt]

%

[wraith turtle]

Thanks!!! Fixed it in this recent update and a lot of other stuff and added multiplayer support lol

[unique kestrel]

and do it on Playstation 2 hardware 🙂

[gloomy girder]

Anyone have a recommendation for starter developments kits on aliexpress? Arduino uno etc.

[cosmic hatch]

depends on what you wanna do

[gloomy girder]

I honestly don’t have a clue.

https://lontentechnology.com/products/custom-diy-custom-robot-kit-motor-car-sensor-for-arduinos-kit-mega-board-pro-mini-nano-for-arduinos-starter-kit-manufacturer

I found this from a Chinese manufacturer, seems like a good start.

[cosmic hatch]

I usually make stuff I wanna use, but if that's your style, go for it

[gloomy girder]

Like custom pcbs etc?

Or what do you mean by that.

[cosmic hatch]

I've been working on stuff similar to this https://www.adafruit.com/product/5128

[keen sorrel]

imo take some time to think through what you'd like to make before buying anything. It's infinitely easier to learn if you actually want to make a thing.

[gloomy girder]

I do have one thing that I want to make. A pill dispenser that dispenses a pill 2x a day.

[cosmic hatch]

do you have a 3d-printer? otherwise it can be janky, but not too hard

[gloomy girder]

I know that would make it alot easier yeah, honestly might get one.

[cosmic hatch]

of course you'll need to make R&D on your own in that case, but the software and hardware for something like that shouldn't be too difficult

[gloomy girder]

Yeah, that’s why I thought it would be a good first project.

[fickle storm]

Hey how are you

[wraith turtle]

that moment when you have way too much fun writing the patch notes that day for some reason 😂

[cosmic hatch]

hello 👋

[tawdry lotus]

what should i know before learning c++ ? (im from c)

[silk eagle]

just start u dont need to know anything

[fading osprey]

Especially not if you already know how to program using C

[blazing mica]

the main different thing is C++ has classes, for OOP

just start learning from now, everything will be okay 🙂

[fickle storm]

I need something to get more files at once of a website

[fading osprey]

Have you considered: ChatGPT, Claude, Gemini?

Perhaps, giving more context would help, instead of taking 4 days to reply, and providing the least useful possible response that is not simply "hi".

Please also consider looking at
https://nohello.net
and
https://dontasktoask.com/

[hollow basalt]

[weak brook]

I love how there's a website dedicated to the no hello one

[fading osprey]

Because it is such a big problem in coding servers is why lol

[weak brook]

ah, i don't hang around too many of them.

most of my life is HDL anyway, so I suppose im not one of the cool kids

[austere bobcat]

I link dontasktoask regularly

[weak brook]

that one i've seen plenty, just not the hello one

[limpid reef]

It's such a big problem in many organizations too. It's not 1980 anymore Dorothy; spit out what you want from me so we can both go home on time.

[frozen flame]

I love small orgs cause you can just walk over to someone and sort things out in a 5 minute conversation

"Hello, please let me know when you have a minute to talk" the time I spent reading this email you just sent me would've been just fuckin fine

[wraith turtle]

Well, another huge update, anyone mind giving it a try for me? Should be immensely more stable ❤️ even has audio now!
https://1jamie.github.io/shape_slayer
I just want to get some feedback on the game so far

[hollow basalt]

No

[hearty pasture]

this is actually pretty engaging
I don't have the time to go over it entirely but I haven't noticed any issues thus far

It could perhaps be a little bit more polished
UNLESS this is the intended style for the game, then you did well on that

[wraith turtle]

The style is intentional but it still needs more polish anyways lol. It's still pretty alpha and the gear system is a place holder but Demonstrates the games concept

It's going to be replaced with a card based system with much heavier meta progression

Thank you so much for your feedback!

[shy ermine]

Does any one know how they handle the phone number in the material ui components library i wanted to know how they verify the phone number based on the phone number length.??

[fading osprey]

https://github.com/google/libphonenumber

[gleaming plume]

Anyone have questions about OpenSSL? I'm in a room with most of the OpenSSL team.

[gleaming plume]

They develop the open-source OpenSSL library and app, which underpin most cryptography on the web.

[prime lark]

Does anyone here have SolidWorks? I found a file online that I'd like to convert to a Zouw step file. Could someone do it for me?

[old arrow]

I probably could but not immediately

[prime lark]

doesn't matter, no hurry

[wraith turtle]

facelift time!!! loving the new look so much!

[mellow path]

Looks good, what programming language did you opt for?

[wraith turtle]

Don't hate me lol, it's vanilla js and canvas 2d 😅

Hey @mellow path XD don't heart break me lmao, I built the whole thing from scratch with no libraries and still get rock solid to your refresh rate fps minus the first 2s it takes to build the sprite cache as the game has zero image assets it's all done with draw calls 😂 give me a little credit

[mellow path]

I'll never hate on a dev

[wraith turtle]

I was more trying to be being sarcastic

[mellow path]

Doesn't matter if you use a markup language to code a project, as long as you made it you have my sticker of approval

[wraith turtle]

both made it and made it run well somehow... Just took 70,000 lines of code

[mellow path]

I personally haven't touched java in like 6 months, and I use JavaScript on the daily

[mellow path]

70k lines 🥀
Id hate to see 70k lines of code in one space

[wraith turtle]

Well it is not just one big file to be fair

It's like 120 files XD

I basically built a whole game engine for my game to run on 😅

[mellow path]

You have a lot more patience than me

Id have a hissy fit and just run unity if I had to compile my own game engine

[wraith turtle]

That was the whole fun of it though for me at least

[mellow path]

Fairs

[wraith turtle]

And hell, now I have something I can build additional games in and it's good enough that if I really polish the gameplay it could easily be $10 on steam

[mellow path]

I wish you good luck

[wraith turtle]

Just wrap it in the ever venerable electron LMAO

[mellow path]

Don't forget me when you become a multi quintillionaire

[wraith turtle]

I'd be happy with a few thousandaire 🤣

Though if you want to try it it's just a GitHub page lol, open source like a weirdo

https://1jamie.github.io/shape_slayer

Only has mouse and keyboard support atm though

[unreal mirage]

i dont even tell my exact problem here lol so,i use typecript react here and i add a menu to navigate some website,but the UI idk what happened it just really suck and i have to ai something but no one works.can someone helps?💀
||my eyesight not good so i posted in a wrong place like 2 times huh||
https://github.com/Kananosa/kananosa-website/tree/beta

[old arrow]

It's your punishment for using a framework instead of straight ES6 like God intended

[wraith turtle]

Holy dependencies Batman

[wraith turtle]

Lost me at vite, my apologies. I've used it once and failed epically and can't speak well for its usability. Quick little cheat I will at least recommend to you if you don't mind I think Google antigravity is in free preview. You might be able to load it up into antigravity and describe the issue and provide a few screen shots it might be able to help sort you out but it is gen ai and everyone feels their own way about it (legit just mean everyone has their own opinion)

[unreal mirage]

ok im gona try that,thanks for advice

wow the installer just like visual studio code

[austere bobcat]

https://tenor.com/view/frozen-kristoff-kristof-somebodys-gotta-tell-him-kristoff-frozen-olaf-gif-27199265

[unreal mirage]

lol every ai ide like cursor just based on vsc lol

[austere bobcat]

Build off the best 😤

[unreal mirage]

Very stable AI i have never used before

[simple canopy]

What is your guys' take on the new "Agentic IDEs"? I.e. VSCode with AI BS strapped to it?

[unreal mirage]

Everything just so laggy and unusable

[cloud knot]

it depends on how much you use AI for. My AI use so far is "hey, explain this error to me" or "i want to do this in a language i never used before, how would the code look like?". But i would never use the results 1:1 without checking first. So for example i needed a powershell script for specific URL download -> XML parsing -> another URL download based on XML results. While i have no Powershell knowledge, the AI made the script, i manually validated and played around with the code etc. And for stuff like code complete etc, i see no much point in AI yet, but that's me and me running only local models. I ain't paying for AI

(other than my electricity bill, that is)

[simple canopy]

See it pretty much the same way

[cloud knot]

and for this i can run the AI within Ollama/LM Studio; or use Continue in VSCode. I haven't really used Cline that much yet for more 'letting AI do stuff' workflow.

[unreal mirage]

i am freaking love gemini

[fading osprey]

I hope that this is satire

[unreal mirage]

The Claude sonnet 4.5 just works so good💀

[cloud knot]

that red sea is usually result of one missing bracket

[boreal barn]

https://www.youtube.com/watch?v=y8ddg4460xc

[west idol]

Ok need help... I mostly code or do D&D stuff in my PC (Ryzen 7 9800X3D with a RX 9070 XT), should I go Nano IPS, OLED, or what other monitor type that is around $700 should I get????

[jolly sable]

I have found cursor composer works relatively well. Sure it’s not great at ui but it’s fast and can get stuff done

[wraith turtle]

i tend to just run on they are nothing more than a tool in your toolset and the better and more precise you can tell them what you want them to do and even how and with what tool it can do some good stuff, but you have to fully know what you are doing for it to produce anything worth while that doesnt have 30000 bugs

it's at least hella good as a templating engine lol

[cloud knot]

or run expensive (remember that guy who got a $120k bill from google), or have extreme security holes (who needs to protect their storage buckets with passwords)...

[wraith turtle]

Well there is always a time and a place for a tool, that is one you don't use it

[keen sorrel]

holy shit, vivado got a new IDE design and finally added an official dark theme. no more manual theming needed. Maybe this was in 2025.1 too but I just hadn't noticed the toggle

seems to be just theme changes but I like it. Dark theme is pleasant on the eyes. No more flash bang switching between vitis and vivado

[keen sorrel]

so nice

[trail belfry]

hello yall, i request some comments and suggesstions on these forms i made for my business(i dont know if this is related to this server, i do think it is cause its about software)

[neon oriole]

as far as i know Tim Berners-Lee had nothing to do with javascript , not frameworked not plain... 😄

[neon oriole]

well it could be if we could check the code behind the forms, 🙂 maybe , this way its more design i think(wich in the end also is (or maybe was 😄 since an interface designed by a real programmer can be spotted from a mile away) a part of programming so

(take a look at screenshots of "bulk image renamer"if you want to know what design by programming engeneer means :D)

[old arrow]

what

[neon oriole]

god -> tim berners-lee , , you know "the creator" . no js in his creation ...,
|| figured that was an easy one to get actually||

[old arrow]

I have no idea what you're trying to say

[neon oriole]

, defenetly the creator, he did html spec the original one, that one has no javascript in it ,... , since he is the creator of the web and he had no javascript , therefore raw ES6 cannot possibly be the way tim berners lee (god) intended

(well maybe tim berners lee is not god ,as he only invented the world wide web, the internet on the otherhand ,was invented in belgium,... (not a joke actually , it really was by a dude named Paul Ottelet or so 😄 )

PS i really think we should stop calling LLM's and other transformer based systems , AI, what are we gonna call AI when someone actually gets around to doing that? especially when everything called AI is not even the whole pacakge as you only get access to the not learning not intelligent part, so just a statistical statemachine

[old flame]

we're going to call it AI

[severe jolt]

I am a just full stack and full time developer who finds freelancing / contractor roles.
Anyone I can help with ?
This is just my portfolio
https://slobodan-stankovic-portfolio.vercel.app/

[neon oriole]

you think ,... marketing divisions all over the world ,.... are gonna go with , just name it the same as the old thing , it will sell without people being pushed to buy the new name thing?

And damn a full stack developer, thought them job offers were wishfull thinking but they do exist 😄 (just making fun of the joboffers here , not you, we need a generalist that specializes in everything ...)

[wraith turtle]

when you manage to do everything on cpu and it still runs like a dream XD (but only because you optimized the daylights out of it)

[old flame]

yes. or AGI if you prefer

[lyric kite]

I feel like I just did something productive with my server for the first time with regards to my discord bot & website integration for said bot. I've worked with Git for school projects before but nothing too serious.

Today I setup a private repository where I can edit my code locally and push it to the repo, and it's automatically pulled into my Proxmox Ubuntu server using coolify. It's nothing much but it's my first time so i'm pretty proud of it, I used to edit my files locally and paste them over via WinSCP then sshing into my server to deploy. Before that I was using the Proxmox built in console to code on my server.

[shrewd valve]

if you want an actual repo to host your code, take a look at gitea. basically github but selfhostable

[peak acorn]

I typically just use github because its all gplv3 stuff anyways, i want it to be shared

unfortunate result of that is that it will be scanned by AI but thats kinda the cost

[crude kraken]

Can someone help me with github pages, I'm trying to make a project but the thing isn't deploying to pages like I want it to

[shrewd valve]

Ask questions about specific problems you are having. Almost nobody will react to such broad help requests

[wraith turtle]

I mostly agree but only because ai is a broad term more colloquially known as machine learning. The proper term is ai but in just the literal sense of a category, however, that's the thing. "LLM" is the proper precise term. The underlying technology is technically ai/ml however even your camera on your phone that the software cleans everything up is also ai, so is facial recognition or object detection or even classification. These are all technically ai/ml which is why I think we need to start directly referring to them as what they are "Large Language Models" as using the general AI term is kinda giving all of ai a bad name because it's polarizing atm with people, along with image diffusers. Ai is just the underlying technology not the full implementation

[lyric kite]

I’m gonna need to look into that, thanks!

[neon oriole]

I really think the proper term is not AI, (even if we call the logic guiding npc's and opponents in games AI) it kind of really isnt AI as per the meaning of the word , since by intelligens we usually think of an entity that is selfgoverning autonomous reasoning desicionmaking (and possibly acting but not a requirement ofc) but even aside from that, it wasnt really my point ...
what we get to use as people is only half of the LLM, we have no access to the backpropagation nor is it actively doing stuff in the background , what we use is more like , if we asked einstein a billion questions , wich he answered and wrote down , and if anyone asked a question now , an automated search system looks trough the questions and presents it, ... we would not call it intelligence but a statemachine ... an llm is allot more complicated but from the point we start communicating with it, it has stopped learning , and there is no possibillity to teach it anything new from our side, ( even if we download the modell, do some training, then "export" it again, but still when we ask it something, its just the exported statemachine that formulates the outputtokens ,. there is no actual "reasoning" or "thinking" going on , or some intelligent, i dont know enough about htis subject let me go look it up , or find myself a paper , analize it myself and then present the user with an answer im confident is accepteble"

what john carmack isdoing atm (claims to have almost figured out what they still lack in understanding and technology in order get started , ....) finally gets going and when it reaches maturity , that is smth i would call AI (even if specific and only fit for a single purpose (so not even AGI yet) it would atleast run more dynamic and produce unpredictable (as in not a statemachine) output

[neon oriole]

as an aside , here where i live there was a Telco/ISP that 20 years orso , started selling an internet subscription under the banner "Fiber", at that point they were upgrading their cable network to have the local distribution hubs be interconnected with fiberoptic , note that the street distributionpoints were still just using coax to the hubs and that the endpoints (users) were still also just using coax to the distribution points... now they are trying to convince their subscribers to go for the "new" fiber , as its allot faster , you can imagine the regular people ( wo got 300mbit down already) declining since: 1:its fast enough , 2:they already have fiber , so theyre fine (with the new fiber connection there is an action that makes the final few meters of fiberoptic free to be installed) , 3: theyre safe from being disconected when the coax network phased out. I think they verry much regret naming it "fiber" whenit wasnt really fiber yet

[gaunt torrent]

Hey guys how well does Cursor AI work for y'all making Android apps? For making an app from scratch, after about 2 prompts I got it to be able to run at least, but with a whole lot of bugs, errors and faulty components, but even after ~10 more prompts I couldn't get another uploadable version and had to use the OG one? Did ts happen for any of u?

[wraith turtle]

I think you make valid points here though at the same time I will point to your specific example, we commonly call the logic for an NPC ai, it's not actually intelligent which is why I feel we call it "artificial" intelligence because it's not actually possessing "intelligence" it's faking it, doing it artificially. We call fake grass "artificial" when it's 100% not grass but it just has a similar usage. It's not true intelligence which is why I think the term being thrown around these days is instead AGI. Though in general we are just debating the semantics of words 😅. I would say though the world has been using the term ai in that context for that use case and definition for quite some time. I'm not sure tbh, just the way my brain logics it

[neon oriole]

ow i always tought of artificial , like an artificial lung or so , its not biological but it does or fullfills the same function as the original biological part, it must not resemble it or look like it, or simulate it (in that case it wouldnt be any good) but it can be a replacement for the real thing . where NPC ai just mimics the real thing such that optically it looks like it but it can in no way replace the actual thing. , This definition also works for the grass btw 🙂 wich is kind of funny , artificial grass , for soccer or whatever can replace the real thing and fullfills the same function 😄
this is carmack's talk on what he's brewing in his basement btw (its a non profit , so its a much less biassed take on whats happening since i think he's both rich enough and a big enough nerd for these things that he doesnt have to sugarcoat the truth in order to protect the marketvalue of the company sinced there is none) if you havent seen it ,it's well worth watching. In my dad's words (a retired life long robotics engeneer) Damn I actually knew all these things or should have realized them but somehow all the hyping on the one hand and the fearmongering on the other made me lose my ability to clearly see where we are and what still has to be figured out and solved ... ) , and hes right both the ones telling us were at the brink of creating a "judgement day" like scenario , and the ones were one step away of Superintelligence making our life and world a true utopia , paint us somewhere on the timeline , where the mission is 90% complete for where we want to be with AI ,.. and if you think about it , both sides their financial healt depends on telling us were further ahead , and not talking about the things we have not even a clue for how to solve them or the fact that there are pieces of the puzzle that were not even aweare of being a part of the puzzle at all 😄 https://www.youtube.com/watch?v=iz9lUMSQBfY

[wraith turtle]

Honestly in this one @neon oriole i think we are both accurate in this case when looking at the roots of the word itself founded in latin. the origin of the word arificial is around the 14th centry and comes from the latin word artificialis which means "belonging to art or craft". This is further derived from artificium meaning handicraft, work of art, or skill. artificium is already itself a blend of ars and facere which basically boils down to made or done by human skill or "handiwork", not same or true to life, is effectively and imitation of the true thing, like a sub for it thogh it does not mean it is 1:1 or even close like in the case of artificial grass. because by the "artificial" meaining a direct stand in artificial grass neither grows nor processes co2 into o2 though we still call it "artificial". Though i will also point out where you directly end up right too as the meaning has changed a bit over time and has started to lean towards a direct drop in like in your case of the "artificial" lung, it does everything you would expect a lung to do and does it properly just in different ways, meanwhile an llm is nothing more than a farce of being "intelligent" when in truth its just really really good at guessing the next word to a scary amount of usefulness, so i think you offer a rather valid point and we start muddying into the actual modern use of the word which does not necessarily directly correlate with its latin roots, words change meanings over time and evolve separate from their roots.
TLDR: i think it falls up to the person as "technically" speaking from the origin and root of the word it just means "something made by man" though at the same time the use and me3aning is starting to skew on the direct meaning so either can and are used interchangably while still being accurate.

sorry, spent some time researching the word before i threw egg on my face 😂 wanted to make sure my rough understanding was actually founded and not just something my brain concocted in the background lol

[neon oriole]

okay did not know that:)

[wraith turtle]

the word artificial i think is a bit too broad tbh, we could probably come up with a more accurate term

like by the origins of it, it just means something made by man, which basically describes 99% of what we interact with every day (bs number for the point, not actual numbers to clarify)

[neon oriole]

given the fact that you actually knew that (or looked it up, same deal) ,... i think you like (||sarcasm||) the recent trend of doublespeak and words no longer meaning what they mean for some of us, about as much as i do,... and then we find ourselfs using a word that has more or less a double meaning and tripping over it ,... found it somehow funny

[wraith turtle]

i actually tend to lean the opposite direction, i like to look at the exact definitions of words and use them as defined by that but i will say in doing so it lands me in some funny situations. my sarcasm detection is crap and i'll be honest lol

[neon oriole]

thats what i meant ->like+sarcasm =dislike

[wraith turtle]

lol, well i guess that just furthered the point lmao

in general if i make a statment before i follow it up if i am not 100% sure i look it up and learn about it. my rule of thumb is one statment is a general knowledge and information i have on hand, if i debate it i need to be 100% sure or i need to research it. I have this obnoxious fear of putting incorrect information into another persons brain as a fact because i didnt truely know before i said anything. to to clarify this is just how i drive myself and deal with information

[neon oriole]

i have this thing where i can't make a choice or whatever without havving enough data to support that choice, also given more data , i have no problem- as im older i tend to stick with my current position untill i had time to process and verify that new data- but i used to if the data was clear and trustworthy ,swichtch my viewpoint in the middle of a discussion,... comes with the downside of the complete inabillity to make a choice if no data is availebel or if its to close to call,... or even worse is down to personal preference ,... if its important like you , im down with researching the crap out of stuff,... wich has gotten harder and harder the past few years, the internet used to be searchable and open for all , but most ofall , once on the internet it would not dissapear, but now i cant find the same documents i found last month , everypage is loaded dynamicly with javascript so good luck scraping(||no its not illegal,... i tend to follow robots.txt rules but even those arent the law , and not following them is not a violation of any law i know of, also i run and share my own YACY server||) a site for all its contents. I dont fear putting wrong information into someones brain, never tought of that, I just like havving the correct information myserlf to base my choices of. Basing it of incorrect info seems counterproductive ,... and i hate doing the same work twice (allot of my projects require a rewrite actually they work but i kind of lost intrest once i had solved the problem they were intended to fix... 😄 , and ive tried to get AI , or better yet LLM's to do that for me but they can't seem to do what i need them to do , DON't change the style , DON't change the naming schemes ,... and for the love of god dont change the indents or alignments ,... 😄

i solved the tabs and spaces thing a few years ago, there is no right way for personal use , but there is really a better one , for storing and sharing code that others need to read, print and or work with... that is if you dont want to complicate things for your fellow humans that is... , and its actually not that hard to figure out whats best for storing , as there are allot of factual arguments for doing it that way,... starting with , how blind people use computers and how many characters they can display on their braile displays,... , to write a parser that can determen the indentation used that is always correct even with multilang files , if you dont want to write the code for the indents your currently using and its variations, and things that migh't appear in files and code that are exceptions to handle , and parse before you can tell with 100% accuracy what the indent of the file is well then prolly thats a good reason its not the better of the two

[old arrow]

Yeah I'm thinking I'll start using tabs in my newer projects

[wraith turtle]

I agree though at the same time memory is not perfect and I can't truly remember everything. I'm in my 30s but full well understand how little I truly know. I just try to always carry that thought with me and everything I recall I try to gauge myself on how confident I am with said information. Some things I can 100% some I can't but I try my best to always be aware and if I don't see it as valuable to research I just disclose it when I state it. Hence my "I'm not sure tbh" in the message. I like to be accurate, clear and conscious of what information I put out there. The brain is not some perfect recall machine and I try my best to live within my limitations. Let's say some info I recall from school, I might be confident in it but that could easily be false confidence, even "common sense" is commonly referred to as "prejudice obtained by age 18". It's not hard to accidentally get wrong information somewhere along the line because someone else did exactly that and presented information that was wrong as facts. Hell even teachers can accidentally do this without meaning to, because they thought they were right. Heck even white papers are retracted every month. It's always good to evolve and grow your knowledge and always be sure to validate things. Not all incorrect information is intentional or even known half the time.

Though as for using LLM for programming I don't have many issues and it does a large majority of what I ask as long as I am detailed and ask it right and make sure to point it to the right files. Usually I will take that and have it draft an "implementation plan" based on my code and reference it and what I am asking, that I read over and adjust as needed to be accurate from the files that need to be modified to what specifically needs to be done and with what pieces. I use it sometimes to tweak and adjust things in my game that is over 70,000 lines of code and it does pretty well, though I will say it's the most powerful for analysis and investigation. Like I can ask it to go through my code and calculate the average numbers for different rooms between enemy scaling and player scaling on average and it does a pretty good rough estimate which is pretty good for game balancing numbers

[wide reef]

What feedback do you expect off vibecoded pages?

[neon oriole]

i mostly user llm's to write --help and manpages for me, it serves as a double check aswell:
code -> what i think i programmed
manpage-> what the llm makes of it, if this is different from what i think i made i need to go check were i went wrong
if the help the manpages and my expectations match chanses are the program does more or less what i want it to do
then i can go and fix the --help ,the llm gave me because for shure it will make it look like a dump of a --help
so far however i cannot come up with a single time i used an llm for writing code i did not regret it hard after a few days, usually doubles the amount of work i have to do.
llm's produce unhinged code that is thrown together . from average quality programmer snippets, wich makes it even worse , at one point it does a branchless solution the next its a ifelse tree, solves 3 nearly identical parts with different approaches , not seeing that if it would have used the same approach it could simplify the code 10fold. for every problem you get it to solve , it adds new code ,it never combines existing code or adapts it , let allong removes code to merge it into the new approach ,
in the end you have a one man project that looks like it was programmed over the past 5 decades , with it being handed over to a new team of devs every week...

if it was any good at programming , they wouldnt be trying to sell it to programmers but they would keep it and sell its output to companies , at half the price of a team and finished in half the time aswell, they only sell it to programmers in order to have them think it can do stuff while theyre the ones fixing and adding data to the thing so perhaps maybe some day it will have enough code to pick from ,...(read , based on what, how often that code appears?) to produce a coherent result

[fading osprey]

Don't always assume because the llm says something which isn't what you understand from your code, that it is your code or in turn yourself which is wrong

I was using an llm to clarify some documentation earlier today and it got so hung up on how to do something that even moving between three different models, and several threads, it kept repeating the same thing, which was fundamentally wrong for every single published version of the package, because I went back to check out of curiosity later on and found out I was so right.

Just because the llm says it, doesn't mean it is correct.

[austere bobcat]

I'm not trying to pick on you specifically, but this is a verbal tick I've noticed in myself and others.

Just because [x] doesn't mean [y]

That's not proper grammar, right?

I wonder how that became a pattern people use.

[old arrow]

what is wrong with it

I don't really see anyone with reasons why it's not proper, other than it just 'sounds bad'

[neon oriole]

bless you , im wy to close to my 40ies to be comfortable 😄

Memory is indeed not perfect, but when it comes to things I’m interested in as I walk through life, if I have a very strong jumping-off
point for what I was doing, I can do a fair bit of pretty accurate recalling of what happened.
I mean that when something happened i have a strong memory for can be anything really
like the time the PC speaker fell off the desk by accident, right on top of the power plug of the PC, turning it off just when he was about to finish the Allied campaign in Red Alert 2. He blamed me for it since my turn on the PC was coming up and now I could finish the Soviet campaign first. This was the summer after it released (we got it for Christmas).
When that happened, I can almost walk through the entire day. I don’t know if it was a Tuesday or a weekend, I don’t know the weather outside, but I do know what I was doing before and after. I’m fairly sure I can pretty accurately remember the exact words that were said.
It’s like: once it’s happened and is in the past, I have no problem recalling it. But oh damn, if it’s in the future — appointments, deadlines, parties, weddings — I don’t know how people do it. If I could choose what to think about, I’d be thinking about adding that event to a calendar, thinking about checking that calendar, thinking about where I put it, thinking about also checking the date in advance and on the day itself, and thinking about checking the clock every 10 minutes when programming. And thinking about seconds every second, or that what felt like 10 minutes to me was actually 2 hours in real time.

Im convinced nobody chooses what he thinks about (i dont) , but some claim they choose what they think and when thats a trick i need to learn 😛 but havent found anyone that actualy could do that when tested

for the thing to be testable , you actually need to really try and do the command after the question,...

got kids? Dont think of your kids ! ,
got a pet? Dont think of your pet,
freedom of choiceis actually considdereingb every option,..
chooose a random animal that is still alive today...
what did you pick? (eg a Tiger)
where did the idea of picking a tiger come from?
because yo said it fairly quickly and given the number of animals alive today , i think its fair to say you did not consider all of them and chose over the entire set of animals
so who selected teh subset you picked tiger frrom , ah your brain did,... but then its the same problem , did your brain consider all animals to createthe suibset ??

so you cant choose hat to think when you think it , one could say I had more to say in what you toughjt about then you your selfdid ,... but i wasnt choosing either so...
your not even actively involved in picking a set of options to pick the thing your not choosing from :p

[neon oriole]

PS i bet there are others here that have problems gaging the people around them , backstabbers , good lor bad intentions. One thing i noticed is that you can bet that people who never had their IQ score tested for real all have high ones(130-180), (where most of them shold have 110 or less,...
when someone is boasting his intelligence. or actually stating his score , more or les ask him or her for how high his iq is... then a short wil e later snipe im with math so simple its doable for 5year olds... but with a nasty brainbreaker in it, smart ones will assume the latter and not say the answer instantly, but most will just blurp out the wrong but obvious answer.
then you point them out that its wrong , most regular smart people will have it at second try , regbular people usually also do , the ones using their fake iq to manipulate or subue others, will have , either wrong and stop swithc context to differnt thiing or how the question is bad, but the best and one i got recently claimed iq of 159:

i walk into a shop and there is a tennisball and a racket for sale , both together cost : $1 and 10cents (#1.10)
The racket costs exactly one dollar more then the ball. What do both items cost seperatley?

actual ansers of that person: ||

• raket 1dollar ball 10 cents
  -> i point out that there is only a 90cent difference between a dollar and 10 cents and it said the racket costa a dollarmore
• Okay stepped into that one , trick question , the racket is $1,10 and the ball is free
  -> point out that that also does not work for the description
  -> then i dont konw amd im not wasting my time on this ...
  || even when catching verry smart people offguard completely and they go with the ||$1 for the racket and $0.10 for the ball|| that they got it wrong nerd snipes them so bad they will have it the second try (because someone tricked their brain and that is verry intresting and nice for i change doesnt happen often so i need to know how they did it :D)

[silk eagle]

i cant figure you out, shadowlandterrorist

[old arrow]

lmfao

[visual bay]

Tennis ball: $0.05 (5 cents)
Racket: $1.05

[weak brook]

this reads like a 2am drunk text

[opaque finch]

Help i think my organization is being stupid again 😑

Instead of migrating legacy google maps apis, for cost cutting they are caching those responses in an already unstable database

[keen sorrel]

This had already achieved timing closure but I found out during yocto machine conf creation that zephyr rtos on the microblaze V cores each need an axi timer. So I added axi timers and now it won't time, how ironic. Hopefully the ML strats just magic away the problem

oh sweet, the ML magic worked

[cloud jay]

Finally decided to try to learn Lit Elements to make a framework-agnostic navbar and footer for my websites.
It took a lot longer than I would've liked to get the ball rolling, but it's really useful lol

Can even setup per-language content.

[hollow basalt]

ok

[neon oriole]

heavy medication might be a cause here ,... should end 12th of januari then

after reading my blurps again , i have no idea why i said it but most of it seems to be trutfully so 🙂

[silk eagle]

its ok shadowlandterrorist, i will always read your blurps even if they are a little bit psychotic at times

[hollow basalt]

^

Keep the blurps coming @neon oriole

[unborn nexus]

Question about the NPM Supply Chain hack, how does one or a company even mitigate that

[fading osprey]

delayed updates, and a general avoidance of the liberal use of other people's packages is usually the way to go

pnpm has a feature that lets you deny upgrades for packages until the version has been published for an amount of time

[unborn nexus]

i mean, what if youre dependent on these packages. do you just redownload them in a new version or scan artifacts for the virus?

[fading osprey]

You download the last known good version and nuke as much as possible of the hardware that ran the code change all your secrets, and start fresh

[silk eagle]

statistically a package suuuurely wont be infected twice so after that it s'all good

[fading osprey]

checks notes mhm, sure, if you want to think that, go for it.

[neon oriole]

also stop using packages that require as much time to learn the api/docs for them as they require to write from scratch, if its not a perfect match dont use it, and if it works freeze the version of the package and just monitor for security fixes in updates, without updating your frozen version. unless its absolutely needed.... get of the "there prolly a package for that" mentallity 🙂 and on to the , "damn this is more work than expected lets hope there is a package for that" mentallity

[unborn nexus]

im talking more from a corporate rather then a personal perspective

[shrewd valve]

if you have infected machines the safest option is nuking the devices and restoring a safe backup of important files in a corporate enviroment

as for updating dependencies, pin the major version so breaking changes most likely dont get automatically pushed. wait a week before updating non security related minor/hotfix updates. update to security updates immediately or also delayed based on for example cvss. run regression tests before rolling out changes. depending on the application use rolling deployment.

major version/breaking changes version updates only manually as they require additional changes in code in most cases

[neon oriole]

could be interpreted as a corporate "culture", was actually written as such , especially the first part, if a small team can write it in a week (a tailored version , where you get total control , and will prolly never require upgrading untill the thing as a whole gets upgraded, makes no sense to , have a depency on , defenetly not as a company unless you want a 3rd party to point the finger at when stuff goes wrong... but in that case ,... no point actually

other than that cant really tell anything more than what is said above, for fixing the thing. But fixing it and keep doing the same will only guarantee , that you will do allot more fixing in the future , so some of the behavior that lead to it has to change in order to not make the same mistake over and over again, as hours spend fixing will soon outnumber ours spend actullay writing the packages you depended on 🙂

[unborn nexus]

We are using package-locks and publish artifacts in a nexus repo
If thats what you mean
But developing all the packages from scratch for ourselfes and not using npm is completely economically unpheasable

I mean you only can check all the artifacts and your Versions for compromise, rigth? That was my question

[neon oriole]

als was more talking about in general , since the day that javascript ecosystems make sense to me is the day i have to check myself into a mental hospital 🙂 companies used to suffer from the "not made here"-syndrome where anything that was not build by the company itself was by default not even looked at as a probable solution, and now its the inverse , as long as we dont have to make it ourself because that might take to long... and unless XZ utils was a unique case, any one package maintained by a single dev that is more or less popular is a possible risk and in the case of XZ we were lucky that microsoft had done a big OOPSIE 2 weeks prior and needed something to distract the media attention at the time so...

[unborn nexus]

Because back in the day was back in the day and an imeasurable amount of things have changed and if youre a multinational corporation hiring an army of devs to Rebuild 1500 packages every year or two is not worth it.

Dumping the entire npm ecosystem is bullshit advise for 99% of Companies

[neon oriole]

never advocated for both, just for the lowhanging fruit that takes less than a week to develop test and publish internally for a small team of 5 or so, and once made like the really obvious ones almost never change , prolly the dumbest example for npm would be stuff like leftpad ,... and i serioiusly hope that 99% of companies dont use npm , but i assume 99% of the onces that already use npm,... but i can only somewhat relate over to python and pip, where a company like ableton heavely uses python for their live product , yet their final product has no depencies on anything comming from pip, and i assume many of the things they require also exist as packages on pypi,
all im advocating for is reducing the total of depencies out of your control especially the ones that dont need vast resources to create and or maintain
afaik most code never really changes anyway

[bronze pendant]

man am I glad to rely on very few deps :3

[old arrow]

you look like you would use rust

[neon oriole]

sometimes you wonder how all those C devs program ey, no packagemanager (well yea cmake maybe but thats only locally on the system), no online repo ,... :p

[old arrow]

there is a package manager: the one on the OS

if you don't have the dependency, you just then install it with your OS package manager

[neon oriole]

yeah sure thats about the same level of user generated packages , similar to npm or pip or cargo ,... , besides they are often the binaries of that code, with header files ,... and verry platform dependand packagemanager ,... for VS you have nuget i guess yes

(its like , hey im develloping a VST, need a lib for a linearphase FIR or IIR filter , let me try with "apt search ....") 😄

[hollow basalt]

Ok

[hollow monolith]

Hey guys, quick question for the other devs here. 👋

I’ve been trying to run a local Anime RAG engine on my Ryzen 3 (integrated graphics potato), and it was obviously choking.

So I wrote a custom Hardware-Aware pipeline in Python that forces Nomic Embeddings (CPU-optimized) by default, but hot-swaps to HyDE if it detects a GPU.

It also generates these Anime Watch History Receipts using Playwright as a visual log.

My Question: I'm trying to sell the source code to fund a proper workstation build (trying to ascend from iGPU hell 💀). Does $79 feel like a fair price for a full FastAPI/Streamlit repo + Scraper? Or is that too steep for an indie tool?

Honest feedback appreciated. Just trying to figure out how to fund the hardware upgrades.

[peak acorn]

Selling software to optimize low end hardware, when there's an equivalent option of buying better hardware, seems questionable

[neon oriole]

you could remove some functionallity offer it as opensource, and have the extra functionallity as fundme/donate plz plugin's maybe , but dunno, people intrested in source code usually can write source code to so you would have to find someone that could write it him/herself but does not have the time to do so but does want to build on the project in the future and is willing to get familiar with a code base he/she didnt write ,... other than that i have no idea what fastapi/... the rest code looks like so cant tell you if 80bucks is allot or not ,...
i might however possibly have another idea, .... if your willing to 1:1 recreate the project, incluse the pitfallse you had and overcame, record the intresting parts , with allot of demotime of partially wroking solutions in between and make a youtube video ~20min of it , in a funny tutorial kind of way but similar to the stuff of sebastian league or so ,

[neon oriole]

i assume that if it runs better on slow hardware , it also runs better on fast hardware , only now you can load allot more on the fast hardware before it comes down , could see a benefit in that for everybody, i optimized the tuithing of usbtop at somepoint, came load , 3times the refresshrate vs the original (mine is the left one) 😄

damnforgot , that it doesnts show on asciinema , because of the frames

[wraith turtle]

lol, got bored a built an nfc based password and totp 2fa system. https://github.com/1Jamie/rfid-pass
i had a rc522 board laying around so i took the lib that went with it and add support for the ntag215 since i had a whole pack laying around from messing with ameibo emulation

even has hardware uid based encryption so you cant just clone a card and use it as the hardware uid wont match

[keen sorrel]

I wonder how I'm supposed to make use of AUX_REF_CLK properly. It's missing from the configurator GUI but the pin(s) do seem to exist. Yet there are two and they're busses??. ig I just have to manually edit the psu_init.c to get around the lack of GUI options but I'm not so sure how to select FPD vs LPD or how a ref clk input could be a bus

Maybe once I look through all the relevant system regs it'll make sense

[weak brook]

I smell a smell. a smelly smell of Zynq ultrascale.

AUX_REF_CLK is a PL-PS fabric clock input. It's an alternate reference clock source for the PS PLL's. In Vivado, it shows up as PS block ports such as: pl_auxrefclkpd[2:0] pl_auxrefclkpd{1:0]
The "bus" is just Vivado grouping the clock inputs into a vector.

How you might use this in psu_init.c is you'll select AUX_REF_CLK as the PLL input source in registers. Each PLL has a PRE_SRC field that selects its reference clock.

[keen sorrel]

Don't forget the + ;)

[weak brook]

tru

might as well include the ™ too while we're at it

AMD Zynq™ Ultrascale+™

I have a ACU15EG on my desk right now lol

[keen sorrel]

This is all info evident from the TRM page and clearly I've found the pins lol, that's in the screenshot. I was busy so I haven't had a look at the reg reference yet but ig there's a aux_ref_clk for every PLL, instead of a shared pin like the block diagram suggests.

[weak brook]

they're not package pins.

the block diagrams is fairly misleading.

it's just AMD being consistent with their diagramming.

[keen sorrel]

never said they were. They're cell pins

[weak brook]

are you actually asking how you should use them correctly or are we just going to argue lol

I can help, but I'll need the part number.

[keen sorrel]

I was wondering why they're a bus, but I've already realized why. You just repeated back to me the information I already found.

Also, why would you need the part number. These are PS pins, they're same in every Zynq US+ device.

[weak brook]

because double checking is always good

doesn't matter if it's "always the same," I always double check against the datasheet.

[keen sorrel]

There is a single TRM for this. It is literally always the same

[weak brook]

part number lets me make sure that I grab the right revision. It's really not that deep.

have a nice night.

[neon oriole]

lets make ourself a nice inline editable , and navigatable datatable for the commandline ,... how hard can it be right i mean visicalc exists since 1980, so should be a breeze right ,....

[wraith turtle]

and goodbye circuitpython hello c/arduino core and the esp32s3 instead of a pi pico
https://github.com/1Jamie/arduino-port
truely do not like circuitpython, almost every time i try to use it i end up back with esp_idf or arduino core because its so wishy washy and unreliable

[wraith turtle]

famous last words when it comes to command line "graphical" stuffs lol. just wait for the dynamic resizes XD

[keen sorrel]

personally I can't stand python in any form. For me, c++ is where it's at for MCU firmware

[hollow basalt]

ok

[tight halo]

Hey dumb question, how easy would it be to switch from the default vivado editor to vim or neovim

[keen sorrel]

pretty straightforward. I don't think many people do much text editing in vivado itself. I can't live without multi-carat editing

ynq MP First Stage Boot Loader
Release 2025.2   Nov 13 2025  -  10:49:34
MultiBootOffset: 0x1F0
Reset Mode      :       System Reset
Platform: Silicon (4.0), Running on A53-0 (64-bit) Processor, Device Name: XCZUxxEG
QSPI 32 bit Boot Mode
FlashID=0x20 0xBB 0x20
PMU Firmware 2025.2     Nov 13 2025   10:49:34
PMU_ROM Version: xpbr-v8.1.0-0
Non authenticated Bitstream download to start now
PL Configuration done successfully
Updating IOPLL to AUX_REF_CLK
PU�%U��5�j
          �k�
             W�������V,W��
                          �
                           �
                            RT�%U��5�j�H�Protection configuration applied
Exit from FSBL

Slight glitch after reconfiguration, but got my IOPLL on a 50MHz reference from an mmcm now. No more 33.333 MHz. All the I/O has nice clean integer clocks. Gone are 99.999001 MHz, 124.998749 MHz, etc. Still debating if I put any of the other PLLs over to 50 MHz too

[tight halo]

Thanks, ive been using notepad++ for vivado, but neovim/emacs for everything else, i wanted to switch it over to one of those two lol

[neon oriole]

ow i got those covered 🙂 ansi escapes im good at , one of the verry verry verry few things i can honestly say i'm better at then the primeagen(when he did his terminal game thing, i was cringing watching what he was doing there , full line coloring by the character, full redraws from 1;1H eg) is 🙂 a quick demo of my redo of the terminal ui of the usbtop (original first) mine as second ,its not the final one but i cant show the latest as it wont build , (but it fills the device names so you can see Daskeyboard 6 : 100 kb/s eg instead of ID3 ... and the busses are colapsible , all done without any library like ncurses just ansi escapes ) what i show in the clip is simpler but it does not redraws (should be obvious because no flickering on resize ) it selectively checs for any value that is greater than 0 and updates that value only , if its 0 for longer than 0.5s it updates it to --,-- so total cpu use is less than the original and updates can be done faster(isnt possible due to how the code was doing the calculations , using pcap and a fixed window as oposed to a sliding one ,.. anyway

hope this one is readable

[wraith turtle]

I'm right there with you, the only reason I used circuit python was ease of prototyping but it turned into a nightmare shortly lol. I usually prefer Arduino or espidf in c++ I just despise python and it's syntax in general... One space off on line 485? Well go screw yourself XD

[wraith turtle]

Ngl as someone who has written way to many terminal scripts and software that is genuinely impressive

The dynamic scaling is next level

[neon oriole]

noe the dynamic scaling was the mistake of me thats the terminal doing it

i just wanted to make the font bigger using ctrl++\

[wraith turtle]

Fair lol.

I was wondering how you were able to tell the scaling on the fly so that makes more sense

[neon oriole]

it increases framesize with the fontsize (wezterm , is otherwise one fo the few terminals with the best ansi escape coverage :))

[wraith turtle]

Afaik it doesn't report back terminal scale so that made much more sense

Was genuinely impressed you managed to handle dynamic scaling for a sec somehow lol

I don't know of any at least off the top of my head with an API or binding for that

Could see ghost doing it maybe

[neon oriole]

wezterm is the only terminal i can think of that correctly processes that ansi code

[wraith turtle]

Im curious how the newer gnome terminal emulator handles it

[neon oriole]

while kitty has its own type of codes that do the same both are unusable since well yeah one is not ansi and the other is verry little supported

[wraith turtle]

The goal is general support in most terminals

[neon oriole]

in my original clip with the table , the slected cell is over and underlined and blinks, in kitty thats , an underline , no overline and no blink 🙁

[wraith turtle]

At that point it's kind of easier just to build your own visual parser and ditch the terminal emulator itself and build your own

Then your users don't have to use a specific terminal

[neon oriole]

the fun things are getting cursor position from ansi escapes , damn , terminal reports that on stdin , not on stdout or anything printf '\x1b[6n'

so in python doing something like : ```py
from subprocess import getoutput
cursorxy=getoutput("printf '\x1b[6n'")

wont work 🙁 since that only records the output on stdout 😄

[neon oriole]

true, so i use it as little as possible and pick things that are generally supported (ignoring the old windows cmd window , the windows terminal is morepleasant and one may expect users that are terminal savvy to use that one i think) but i do think its a good indication of what the developpers of the terminal care about, since the big text , double size in both widht and length (you can also only display the top half or the bottom half or , double widht but not double hight eg...
all these require consideration from the start if you will so if you make a terminal as you would and are basicly done at the point your implementing the ansi codes you have to start over or do some fugly hacks to make that work 🙂

[wraith turtle]

That's why at a certain point it's easier to build your own UI and just plug the backend into a terminal instance lol

I've done it several times with electron xD

Yes I know, eww electron but it's just so easy lol

I do my best to avoid directly using the terminal for anything graphical out of not wanting the pain

Though have serious respect for those who do it anyways due to knowing how much of a pita it is

[neon oriole]

here is an inline menu thingy i made at some point , uses in place replacements instead of redraws to, my slow python code looks more performant than i think he(the primeagen) was using zig at the time's- compiled code... just check terminal size for changes every loop before drawing anything, if what you have draws so far is out of bounds do a full redraw else inplace clear and overwite ,... not that crazy amount of extra work for a much neater user experience

that one reads contents of folders only when you open the folder , so if a new file appears while running it would account for it , else it would be easy as i could account for everything in advance 😄

[wraith turtle]

I bet you could probably monitor the output of the specific target terminal and have some fun

Like with gnome and monitoring the dbus for terminal window changes

[neon oriole]

ah you can just query the terminal 🙂

[wraith turtle]

Didn't know that, honestly I've always treated the terminal as just a raw output systems lol

Just always worked within those parameters

Just smart choices and trying to clear before writing a new frame

At least always means its compatible with any terminal

[wraith turtle]

I usually use bash

[neon oriole]

thats bash 🙂

[wraith turtle]

That at least explains the discrepancy in understanding lol

[neon oriole]

but bash is a shell , not a terminal , but bash can query the terminal to so 🙂

[wraith turtle]

Tbf it's both a shell and a scripting language

[neon oriole]

thewindowy thing is a terminal emulator (used to be a physical machine somewhere connected to the mainframe) the thing that runs on the mainfraime would be a shell, like bash and the terminal would display the shell 🙂 now it has become a seemingly single application but its not realy , yoiu can still run different shell s in a single terminal emulator 🙂

[wraith turtle]

That's why we call them "terminal emulators"

Though I guess you could still easily make one with an mcu and Jack it into a tty on a machine

[neon oriole]

on top bash , below ion

ansi escapes is a way to communicate stuff from the shell to the emulator and back 🙂

like this needs to be in that color or or move the cursor , to this position or how many lines and collums can you display eg

so its not bash doing the coloring of the text but the actual emulator since bash still just sees the escapesequence 🙂

(ow and yes i know you can use \x1b[31m for red to , but that way i can just do rgb colors wich is usually what you want anyway so for the few colors you can do using 30-39 its really not worth it imho 🙂

worst thing about all this is that AI knows all this but if yo dont know that it exists , it will never use it in code ,.. as most ansi examples out there also either dont know most of this stuff , and or refuse to use it in an optimized way , every terminal app flickers on resize because redraws , instead of just inplace updates , so when AI informs you and you have no reason to believe it can be done better there is no reason to go and look it up so all terminal apps in the future will be a flickery slow and unoptimized mess

one of the upsides to AI i recently discovered is that finally there seems to come an end to javascript-heavy webpages that dynamicly load text and content in a single page , as thats not to easy to present in a easy digestable way to AI bots ,.. while the actual html is verry structured and almost tailored to how AI understands and can parse the text 🙂

[delicate igloo]

I was a game developer and i now hate it, sometimes when i play a game i can only think about the technical aspect of it and i cant have fun

[wraith turtle]

For me that's what makes games more impressive imo, playing something and seeing where the dev figured out how to solve a problem in a round about way or the feature you can tell they poured their all into it. It makes me personally appreciate them even more.

[wraith turtle]

I'll have to do some research into it, love hidden features ❤️ especially when it can make things look nicer

Though rapidly finding fedora universal blue horrible for embedded engineering development

[neon oriole]

there is a github page that gives the most common used ones that everyone points to, but in this case the best resource i can come up with is the actual wikipedia page , it lists the most i know about , just not the tricks you can use to , but its a good start https://en.wikipedia.org/wiki/ANSI_escape_code

[delicate igloo]

yeah i kinda get that too

[wraith turtle]

Horribly the only user I can get to properly reach /dev/ttyACM0 is root and adding the user to the dialout was wiped on log out and back in

And due to that accessing it from distrobox is basically impossible

Don't get me wrong I love every other aspect of the atomic system but we will need to come to with better ways to handle cases like this

[neon oriole]

here is one of the tricks i know : copy paste this in your terminal to see what it does , ill disect every part of it as its a bit long but most of it is just to demonstrate whats going on and not the actual trick : ```
printf '\n\x1b[2J\x1b[1;1H\x1b[48;2;128;128;128m\x1b[K\x1b[%s;1H\x1b[48;2;128;128;128m\x1b[K\x1b[1;2H\x1b[m\n' $LINES

[wraith turtle]

I love me a good printf I use it in bash all the time with sed and awk

Just didn't realize it was technically an escape, assumed it was just part of the printf setup lol. Downsides of being self taught, you make some incorrect assumptions

[neon oriole]

\x1b[2J                  : clears the screen
\x1b[1;1H                : puts the cursor in the top left of the screen
\x1b[48;2;128;128;128m   : sets the background color to gray to be used by every character printed that follows
-- this is the trick normally you would have to print a space or a char for the full lenght of the terminal to color the whole line at the top
\x1b[K                   : normally clears the line , but in this case its a quick way to color the background of the whole line
\x1b[%s;1H               : %s gets filled with $LINES so it puts the cursor at the botom left of the screen
-- same trick
\x1b[48;2;128;128;128m\x1b[K : colors the background of the bottomline gray
\x1b[1;2H\                : paces the cursor at the secondline on the screen left side
\x1b[m                   : removes all markup that is currently active and sets it back to default
```  end and the PS1 gets printed to the second line now

printf is just used to print the text , as echo wont work here 🙂

in python or so you can just use print('\x1b[....') or whatever but echo \x1b[.... mangles it up and wont show the escapes,

the result should look like this in a terminal of the whole line :

most of the time when you find code to do that ,it either has a loop to print spaces for the entire widht of the screen each time , or ive seen loops of spaces to first fill the whole of the screen white , and next to fill the whole of the screen but for the first and last lines with black again,... wich end up as a slow mess and cpu hog as you can imagine but try and find a place that sais you can use the clear line escape after setting a color to color the whole of the line, (or even clear left or right of the cursor position to color that part of the line)

if you know what you are looking for and ask it the right way google actually returns a usable result now (stackoverflow not the top answer but its there: (for 5 years already 😛 )

ow its actually the top answer theres just allot fo comments on the question itself that push the answer further down

[neon oriole]

out of curiosity i figured maybe chatgpt also knows and uses this trick so my complaints are no longer valid , but this is what it gave me:
as you can see it uses text.center(width), wich does the spaceshere centered spaceshere , this means your printing allot ofcharacters wich is slow as f for terminals , even if you use hardware acellerated ones it still has to do every character , if you build a whole tui screen like this its a complete shitshow...

as example how i would tackle that problem allong with some other niceties : ```py
import os

def bartext(text,rgb):
tstart = sum(divmod(os.get_terminal_size()[0], 2)) - sum(divmod(len(text), 2))
return '\x1b[{tstart}G{text}'.format(tstart=tstart,text=colortext(text,rgb=rgb))

def colortext(text,rgb):
return '\x1b[1;38;2;{RGB}m{text}\x1b[m'.format(RGB=';'.join([str(i)for i in rgb]),text=text)

def TuiBar(linenofn):
def tuibar():
LINE='\x1b[{LINE};1H'.format(LINE=linenofn())
GRAY_BG='\x1b[48;2;192;192;192m'
tpl='{LINE}{GRAY}\x1b[K{TEXT}\x1b[2;1H'
return tpl.format(LINE=LINE,GRAY=GRAY_BG,TEXT='{TEXT}',)
return tuibar
def main():
draw = True
topbar=TuiBar(linenofn=(lambda :1))
botbar=TuiBar(linenofn=(lambda : os.get_terminal_size()[0]-1))
sizes = os.get_terminal_size()
counter=0
while True:
if sizes != os.get_terminal_size():
sizes = os.get_terminal_size()
draw=True
if draw:
counter+=1
txtheader='HEADER redraws: {C}'.format(C=counter)
print('\x1b[2J')
print(topbar().format(TEXT=bartext(txtheader,rgb=(255,0,0)),end='',flush=True))
print(botbar().format(TEXT=bartext('fOOTER',rgb=(0,255,0)),end='',flush=True))
print('start text',end='\n\n',flush=True)
draw=False
if name == "main":
main()

if you copy the code and run it in a terminal window , try resizing the window to 🙂

or in konsole wich seems to be a bit more responsive on resizing

[neon oriole]

if you ever have the time its worth playing with the things here 🙂

[fickle hatch]

What OS is that?

[keen sorrel]

udev rules behave the same on fedora silverblue and the simplest would be to just give it the devil's number all users/groups read/write. Example for FTDI (0x0403):

ACTION=="add", ATTRS{idVendor}=="0403", MODE:="666"

Being more targeted than just using VID may require multiple rules depending on the device, but you can get the info about device's entire tree with udevadm info --tree /dev/ttyACM0

[wraith turtle]

I did try udev and it still told me to piss off

I used to do phone os dev work once upon a time so I tried that first

It's kinda whatever though, if all I'm doing is offline flashing as root it's not the end of the world

Be different if it was messing with stuff

Actually makes me wonder if I still have my ffos udev rule script stored somewhere XD

That goes back a ways 😅

I enjoy knowing at least one company used it to build a product it still sells lol

[keen sorrel]

udev can be funky but it is the same on all distros, atomic or not. It is the thing that controls /dev device access and set their perms. You also have to do udevadm control --reload && udevadm trigger and I've always to replug the device anyways even though trigger should in theory resolve that. If it wasn't working then your rule was misconfigured.

[wraith turtle]

I'll give it another go when I'm back at my desk

[keen sorrel]

try the loosest possible rule like idVendor only

[wraith turtle]

Fair, should just be able to target espressif instead of the s3 specifically

Or well it's USB controller lol

[keen sorrel]

Don't even target a specific subsystem

[wraith turtle]

It's got both jtag and otg so I'll have to make sure to account for the vendors of those in case it's different from espressif

[fickle hatch]

jtag, xbox 360?

[wraith turtle]

XD

Jtag as in the debugger interface

[keen sorrel]

like this is the real rule from Xilinx. Though they also disable the chip's interface 0 right before that rule because that one does proprietary jtag for xsdb/vivado hw_server and interface 1 is the textual uart:

# Unbind default ftdi_sio kernel driver
ATTRS{idVendor}=="0403", ATTRS{bInterfaceNumber}=="00",\
PROGRAM="/bin/sh -c '\
    echo -n $id:1.0 > /sys/bus/usb/drivers/ftdi_sio/unbind;\
    echo -n $id:1.1 > /sys/bus/usb/drivers/ftdi_sio/unbind\
'"
# Create "/dev" entries for Xilinx device's with read and write
# permission granted to all users.
ACTION=="add", ATTRS{idVendor}=="0403", MODE:="666"

[wraith turtle]

I'll jab it when I get back to my desk, I'm chilling watching some dr who at the moment

[neon oriole]

was that aimed at me?

ps some other terminal fx 😄

but its just linux with kde (gentoo as distro)

usb is a mess if you try to readinfo from the devices themselfs ,... this is what i got from mine , reading directly using libusb1.0 , and vendors just fill whatever they like wherever they like if they fill anything at all ,...

was trying to find a way to get device names in a consistent manner , not doable lol

[keen sorrel]

whelp, time for DFX. rebooting, reflashing, and rebooting with every edit to a vitis kernel is not a good flow lol

[keen sorrel]

still haven't got timing closure, but almost got DFX ready for vitis

[keen sorrel]

hmm, clearly I'm doing something fundamentally wrong with hooking up the coupler

[keen sorrel]

basing it on a versal tutorial is maybe bad. maybe those regslices are really there for an SLR crossing or something

oh, I'm missing the smart connect on static side, maybe those are required. I was hoping to avoid the resource use without them, but ig I can't do that

[neon oriole]

Damn just tried coding with copilot as some sort of autocomplete , and as some vibecoding assistant, needed a tool i didnt want to code myself boring stuff but it had to deal with other code that had platform dependant sections (some files were for windows , others for linux , and the tool was supposed to recognise what platform it was for (verry simple because most of the time the platform was in the name x64-winnt-filename.py or in the parents folder s name like that ,... o my god what a ffing pain , and it was python so something AI is supposed to be good at ,...
every time it had to detect the platform of a thing :(im not kidding its one of the things it created verbatum(i changed the name of the binaries only) and theyre all like this:
note that its a function for package so always a folder not a file , also the path is passed in by us in the calling function as a Path object.```py
def binpath_from_package(package_path):
package_path=Path(package_path)
if sys.platform.startswith('win'):
exe=package_path/'win'/'runner.exe'
else:
exe=package_path/'bin'/'runner'
return str(exe.resolve())

and its the worst/poorest python code, to get a path out where you know the specific path of the two options , and your using pathlib already i can think of do pathlib first then use  ```py
def binpath_from_pkg(pkgpath):
    #assume its a path object as we pass it in like one and nobody els is using the function
    exe={
      'winnt': pkgpath/'win/runner.exe',
      'posix': pkgpath/'bin/runner'
    }.get(pkgpath.name.split('-')[1])
    return exe
``` or something  working atleast would have been nice lol , yeey to the future...

needed to vent a little fustration, as im now writing the code i didnt want to write in the first place anyway 🙁

[fickle hatch]

Do not use ai for coding

If you want to learn

[neon oriole]

problem is that AI has all the information to write that second part , but somehow statistically its not the highest scoring tokens and it wont get used , wich is a problem they should solve before ai is usefull or any code it writes will never be the optimal code for anything , as optimal code will never be the most present in training data , as for every discussion about some code , allot of not optimal code is posted for usually one optimal that on stackoverflow gets the most upvotes but is only a single piece of trainingdata ,...

[neon oriole]

lol 🙂

i'll keep that in mind , but at the moment im starting to forget stuff at a faster rate that i knew then im capable of learning new things still\ 😮

[silk eagle]

sometimes, shadowlandterrorist, sometimes you drop a message that explains everything so perfectly

[neon oriole]

so perfectly ... unreadable and full of typo's and grammar mistakes that most need an AI to decypher it in the first place but ... i'm flattered 🙂

[neon oriole]

🙂

[neon oriole]

okay the thing can now convert existing repositories, still dealing with the github repo creation thing but thats for tomorrow or so , still just interacts like the regular python venv, also is setup for setuptools-scm while at it , and you can switch whatever pythonversion(and acompanied venv) you want even to windows ones, they will be autodetected or can be created to if they are availeble on the system (and for windows venvs on linux you have to have a wineprefix active with the python version installed ofc but ...anyway)

once i get the github thingy done , its also availeble for downlaod and contributions/corrections on my github 🙂 , currently only has 2 extra depencies, being :
click (a cli thing) and Clict (another thing of mine that lets you create python dictionaries with keys assigneble and accessible in dot notation 😄 mydict.group.section.key.subkey="value" eg its on the pypi so should just be installeble with pip aswell 😄

also somehow the damn gif wont display all colors dunno why:

[soft fern]

Gif use a 256 color pallet so however its being converted has to pick a set to use is why gradients typically look bad in gifs

[hollow basalt]

Ok

[neon oriole]

not that it matters that much but dont think gif is the curlpit here as it does do the other blues correctly and they are mathematicly closer to the cyan then plain white is so :S

[neon oriole]

nice thing about it is that if you switch venves , without requiring interaction IDE's like pycharm switch on the fly aswell 🙂

[keen sorrel]

gotta love a non-descript error, but I guess that's what I setup kgdb for
create_overlay: Failed to create overlay (err=-2)

[keen sorrel]

let's fucking goooooo. I can't help but feel that zynq ultrascale+ has been somewhat left behind for Versal after working around so much to get a Vitis extensible DFX-based platform with a static shell. The docs make it look so straightforward on versal

the design that finally timed based roughly on the kria-dfx-hw project (without SIHA tho)

[keen sorrel]

huh, turns out the reason that sdtgen wasn't handling intc inside the RP was because there's a memory leak when the driver is unloaded/reloaded. Haven't actually confirmed that myself yet, but I just read that in the versal guide

[neon oriole]

i really hope RAD debugger's next step (linux support) doesnt take that long anymore , looks like a debugger that is more or less nice to use

[keen sorrel]

still proving challenging to get the loading of an xclbin to not hang the whole system

[neon oriole]

time well spent : 😄

it still has a bug because it actually remembers its locations so for the top and bottom lines it only remembers the last char as a location 🙁

[opaque finch]

Bros

This company I applied to is asking me if I am available 7am to 4 pm for early shift.

I kinda like to workout from 6:30 am to 7:45 am.
What to do ? Say yes or no ? I have a job now i'm just looking for a change.

[frozen flame]

Why on earth would you not move your gym schedule if you're being offered a job you're interested in

I'll never understand gym bros

[old arrow]

and what does it have to do with development

[keen sorrel]

never been so happy to see a kernel panic. at least it's something more to work with compared to a hang.

[  381.098241] Internal error: synchronous external abort: 0000000096000010 [#1] SMP
[  381.105734] Modules linked in: cfg80211 al5e(O) al5d(O) allegro(O) zocl(O) xlnx_vcu uio_pdrv_genirq openvswitch nsh nf_nat dmaproxy(O)
[  381.117850] CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Tainted: G           O       6.12.40-xilinx-g31626ef92ff1 #1
[  381.127761] Tainted: [O]=OOT_MODULE
[  381.131240] Hardware name: ZynqMP KR260 revB (DT)
[  381.135937] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[  381.142888] pc : xintc_read+0x1c/0x50
[  381.146551] lr : xil_intc_irq_handler+0x44/0xb8
...
[  381.225731] Call trace:
[  381.228170]  xintc_read+0x1c/0x50
[  381.231477]  handle_irq_desc+0x40/0x58
[  381.235217]  generic_handle_domain_irq+0x1c/0x28
[  381.239826]  gic_handle_irq+0x68/0x98
[  381.243481]  call_on_irq_stack+0x24/0x4c
[  381.247395]  do_interrupt_handler+0x80/0x84
[  381.251571]  el1_interrupt+0x34/0x54
[  381.255138]  el1h_64_irq_handler+0x18/0x24
[  381.259226]  el1h_64_irq+0x64/0x68
[  381.262620]  default_idle_call+0x54/0xc0
[  381.266535]  do_idle+0xa0/0xf0
[  381.269590]  cpu_startup_entry+0x38/0x3c
[  381.273505]  kernel_init+0x0/0x1d8
[  381.276899]  start_kernel+0x564/0x704
[  381.280553]  __primary_switched+0x80/0x88
[  381.284558] Code: b948c842 8b010000 7100005f 540000ec (b9400000) 
[  381.290641] ---[ end trace 0000000000000000 ]---
[  381.295250] Kernel panic - not syncing: synchronous external abort: Fatal exception in interrupt
[  381.304025] SMP: stopping secondary CPUs
[  381.307945] Kernel Offset: disabled
[  381.311420] CPU features: 0x00,00000080,00200000,0200420b
[  381.316810] Memory Limit: none
[  381.319858] ---[ end Kernel panic - not syncing: synchronous external abort: Fatal exception in interrupt ]---

though I'm not sure how much value it has considering I just programmed the vitis kernel's partial bitstream from vivado without decoupling the RP

[keen sorrel]

Damn, I'm stumped. I even switched the iopll ref back to pss_ref_clk and removed the RP's intc to have the concat directly drive the GIC port. Loading the xclbin even locks up the PS debug port so I can't review the system state through that. What's additionally strange is that the ILA I put in the kernel by enabling debug on the AXI input port only shows up when I program the vitis produced bitstream via vivado, but not when I load the xclbin or use fpgautil on the bitstream extracted from the xclbin. I've manually configured the probes file but it still complains about the ILA being missing.

ig I could add the ILA on the static shell side's axi smartconnect instead, but I shouldn't have to do that. At least it's something else to try though

[keen sorrel]

oh, vitis's GUI for platform creation doesn't handle DFX platforms. Found there's additional args available in the python API for it from the vck190_base_dfx platform on github

hopefully it'll be the last thing I have to fix but I'm not particularly optimistic anymore lol

[keen sorrel]

oh shit I'm actually getting an error code returned when the xclbin is loaded, but it's still locking up the whole system lol

[old arrow]

https://www.scattered-thoughts.net/writing/how-safe-is-zig/

[stark geode]

Zig? What's zig?

[fading osprey]

a programming language

https://ziglang.org/

[neon oriole]

one that is compatible with C , and can interact directly with it, you can convert between C and zig and back , its compiler can also compile C and C++ (llvm ,... but still) and the language overuses the word const,... if you grep zig code for the word you will be baffeled,... (it kind of bothers me in the way that its a word that does not tell me anything about the logic and its everywhere making it harder for me to figure out the actual code , other people dont seem to be botherd by it.. if i had to pick one zig or rust ,... i would still pick zig as rust is just not for me 🙂

[fading osprey]

The crab will come for you one day, just not today...

[silk eagle]

i like rust crab

[old arrow]

The article goes into more than just rust vs zig BTW, I shared because it seemed thoughtful

[neon oriole]

Im fairly sure it wont, becaue you see, i dont hate rust or anything, I can see its value, for certain fields of programming , i can also see that it has one of the better typing systems in existence. and all the praises , however its so orthogonal to how i write code in praktice that it would fustrate the hell out of me , with verry little prospect of ever becomming proficient in it,... so its a purely subjective "Nope Not For Me" ,. but just to not be all positive here , here is what i like the least OBjectively about rust,... (i'm gonna get flak for this i know :P) : the Cargo /Crates thing... starts to feel allot like NPM javascript behavior,... and i dont think of it as particulary secure, so what good is a memorysafe language as the supplyline might be contaminated 😛 (here is an ebuild of a rust program , if you open it you can see what i mean 🙂 and this is jsut the first one i grabbed ... https://dpaste.com/E2QS3SZNL

[neon oriole]

Intresting point of view in that link btw , I never tought of zig as a safe language (think it was invented for the purpose of making a DAW, so security might not have been in it as a core principle) So i think that zig suffers from havving been called safer than C, and it becomming "as safe as rust" due to a game of telephone. I its easier to work with and programwhith as rust , odds are that its a less secure language (or that it has a GC ofc),... but i dont think a language with no restrictions and as free as (old) C that also is safe , can not exist by its verry nature , the restrictions make rust safe(r) and working with it esponentially harder 🙂

on that note also i found this one of the most intresting talks that tought me more than i would have liked 🙂 ow and no its not a takedown of OOP 🙂 its just telling and showing that yor doing it wrong (:P no actually i mean) or atleast worth it to take a look at :https://www.youtube.com/watch?v=wo84LFzx5nI

instead of creating persons classes you create classes for the data like names , health , or whatever , and have the peson just be an ID instead of reading a property of the personclass you query the value of ID in the health objects. ``` nevetheless worth watching 🙂

[frosty hemlock]

@wide reef Hey, when will you be adding new features to billard?

[wide reef]

Put it on break for a sec, new year preparations and exams

currently focusing on a e2ee chat app

[frosty hemlock]

Ah understandable

[frosty hemlock]

Nice

[wide reef]

You can try out the latest deployed version here http://78.81.146.12:666/

[frozen flame]

Got ur ip

[neon oriole]

sometimes i wonder how i end up with these things 😛 : ```py
ab=[*str(sum([*map(int,str(f'{a},-b}').replace('.','').split(','))]))]
ab.insert(str(sum([a,-b])).index('.'),'.')
ab=float(''.join(ab))

prolly should participate here sometime, since my regular nonobfuscated code tends to be close already

https://www.ioccc.org/

[frosty hemlock]

Im working on something cool too now

There are probably already a ton of these solutions but still, for me its about learning new stuff:
https://github.com/JordyDevrix/upback

[keen sorrel]

highly recommend using the coresight dap over jtag instead of kgdboc if anyone is ever trying to debug a kernel on an embedded arm device. It doesn't gracefully handle multi-core, but actually functioning at boot-time, not relying on the kernel working, and being able to halt execution without requiring sysrq g makes it way more useful. Just had to script a few things for pausing and continuing

[fierce gazelle]

blackjack game I made using tkinter python library

[spiral moss]

cool now make it so you can use your credit card and actually gamble and make bank

[fierce gazelle]

this is a very good idea

[keen sorrel]

https://github.com/Xilinx/system-device-tree-xlnx/blob/master/zocl/data/zocl.tcl#L55
this confuses me. Is that not supposed to be decimal?? Also assuming irq0 seems a bit questionable.
https://github.com/Xilinx/system-device-tree-xlnx/blob/master/device_tree/data/common_proc.tcl#L7319

Neither sequence even matches the TRM, but maybe there's some offset introduced somewhere (maybe imux?) that makes this make sense

oh yeah, imux offsets by 0x17 so that actually works, nvm lmao. I'm still confused

still an issue of hexadecimal vs decimal in zocl.tcl though

Oh the mapping of interrupts just isn't sequential

that's confusing af

[white escarp]

i don't know if this is the right place to post but i created an open source istats alternative if anyone wants to use it or even contribute to it

https://github.com/sarzz2/macStats

[stone marlin]

I'm trying to identify hostnames on a local network via the ARP table using Powershell and CMD. Im reading address via the ARP table, and then passing them to various methods to try and get the hostname. Various methods of doing this are being VERY SLOW (>1s response time). I'm trying various methods to speed this up like using regex to parse ping results.

    $address = $_.Address
    $hostname = "Could not retrieve host name"
    Write-Output "Testing $address, hostname currently $hostname"
    
    $test = (ping -w 500 -a -n 1 $address).Split([Environment]::NewLine)[1] -Match "Pinging (?<hostname>.*) \[$address"
    if($test) {
        $hostname = $matches['hostname']
    }

    Write-Output "$address = $hostname"

Trying to use the waitout parameter seems to always default to 4s. The count parameter does have an effect. Is there any reason why you can't specify the waitout time in powershell?

Requirements for other suggestions:
"Low" result time for any command
As backward compatible going back to Win7 as possible

[keen sorrel]

Input image height : 2160
Input image width  : 3840
Found Platform
Platform Name: Xilinx
Input Image Bit Depth:8
Input Image Channels:1
NPPC:1
XCLBIN File Name: krnl_sobel
INFO: Importing ./krnl_sobel.xclbin
Loading: './krnl_sobel.xclbin'
before kernel .... !!!
after kernel .... !!!
526.864ms
    Minimum error in intensity = 0
    Maximum error in intensity = 0
    Percentage of pixels above error threshold = 0
    Minimum error in intensity = 0
    Maximum error in intensity = 0
    Percentage of pixels above error threshold = 0
Test Passed .... !!!
terminate called without an active exception

It only took me nearly 3 weeks, but I finally got a DFX vitis platform working on my k26. Gonna clean it up a bit before making a template project so others don't have to wade through that barely documented experience. I can finally get to learning Vitis...

[old arrow]

https://vxtwitter.com/i/status/2008210510605603017

Travis Whitaker (hs/acc) (@TravisMWhitaker)

“just pick anything not representable with types, e.g.”

> something representable with types

QRT: abstractalgo
@kettanaito just pick anything that's not representable with types, e.g. `div` function whose divider param shouldn't be a zero.

for everything that's representable with types, tests aren't needed per se.

vxTwitter / fixvx · Jan 5, 2026, 4:14 PM

I've been watching all the type theory arguments going on on twitter

[spare dove]

what os is good for development now?

[frozen flame]

Development of what

All of the major ones are fine

But some are better depending on what you're doing

[spare dove]

c and rust

and just basic tui stuff

[frozen flame]

Rust is equal on whatever, but c can be a bit of a bitch on windows I'm told

[spare dove]

yeah windows is a no go for me in general

[frozen flame]

Unless you have a Mac it sounds like you're settled on leenux then

[old arrow]

C on windows depends if you're trying to write stuff targeting Windows or not

[frozen flame]

Or are you asking for distros

[spare dove]

bsd or linux or unix idm just any os thats good

[old arrow]

I believe if you're writing windows programs on windows, C is fine

[spare dove]

windows doesnt boot war thunder without bsod

[frozen flame]

it Really Doesn't Matter which distro you're on. Pick one you like using

Mint is usually a decent desktop

[old arrow]

That hardly seems relevant

[spare dove]

yeah but its still a non negotiable for me how can my 2018 phone do something windows cant

[spare dove]

ill prob go for arch or gentoo then

[frozen flame]

Gentoo is a crazy choice

But whatever floats your goat

[old arrow]

arch has been a very stable distro in my experience (especially compared to Ubuntu and family) but it depends how much you're willing to learn and know what you're doing

[spare dove]

i know alot about arch and the aur and pkg builds

[neon oriole]

or stuff completely not reliant on the platform (i wrote a terminal tool that only needed text input and produced text output, and i used the zig cc compiler to create a linux binary and a windows binary wich both worked great (after adjusting some memory allocation from the stack to the heap as windows only gives you 1mb stack by default

i run gentoo as a daily driver on my main rig , its far from perfect but the least opinionated distro around that is still a distro imho 🙂

[weak rain]

Arch and Fedora are good ones.

[wide reef]

Are you fucking 12

[wide reef]

How’s the project going?

[spare dove]

?

[neon oriole]

also made a project (in python) for terminal interaction , is not for setting colors eg libs enough to do that and ansi is not that hard , but it should simplify reading characters from stdin as they are pressed, tracking terminal colors , size, tracking moving the cursor and storing cursor positions for later recovery , eg ... project can be found here there is a folder with some examples for how to use the lib (dont get the release as its not the latest , just clone the repo add a venv to it and install the repo into the venv 🙂 (its also on pypi but that one does not include the examples) https://github.com/hoefkensj/libTerm

[old arrow]

I like the contrast

[frosty hemlock]

Great

Got a lot of functionality working already

[silk eagle]

cool font what font is that

[neon oriole]

using uuid's still seems like a weird thing to do, imho espcially for backups , you want them to be unique but (this is not one i use for backups but for tracking history of smth) i usually think of smth that is bot meaning full to what it is and unique for that thing liek : ```sh
$(uptime -s | tr -d '-: ').$(hostname -f).#PIDOFPROCESSHERE

[frosty hemlock]

Jetbrains mono space

[frosty hemlock]

The backup application uses a database for keeping track of the backups, every backup has a timestamp, UUID and filepath

So I can sort on all of these. I made the backups file based first and stored the metadata in the filename but than I changed it to a db

I just need to show a not found state in case the file can't be found for some reason

[neon oriole]

ah the point that i was trying to make is that uuid's only purpose is having a unique identifier... stores no further information so i prefer using an identifier that does store some information , but still is unique for that usecase , eg for the history thing i did , using the time the computer booted +fqdn + the pid , is like a unique identifier. but sorting on it also sorts them chronologically since the pid only goes up once booted unless your running so much processes and or have an uptime long enough to hve the pid counter flip around back to the start .... advantage of using that as identifier is that they are verry easely comparable radable even memorisable , wich is verry hard for a human to do with uuid's. downside is that you have to make sure the combination of data to use covers every aspect of the problem for making it unique, if it only ran on one system i could drop the fqdn in it but since i aggregate over multiple systems there is a possibillity that two systems get started at exactly the same time so , i had to add the fqdn to it to be sure, but i always ty to add what thing that causes the possible conflict so ...

if i was storing facebook posts , the system falls appart since the data needed to make it unique would be of a crazy lenght its no longer worth it , and thats when i would use uuid's

another possible downside is that you have to store the alog you use to create the id ,in the database aswell, and you have to take into account that you might need a version two of that algo or more

[hollow basalt]

Ok

[frosty hemlock]

Mhm yeah, for me one reason I went from using unique metadata as identifier to UUID's was also kinda because of SRP, where a method should only be doing 1 thing and 1 thing only. Using ur identifier as identifier and as metadata would kinda break SRP

I might be too strict in this, maybe not maybe yes, idk

[frosty hemlock]

What are the 3 icons under the messages btw?

[wide reef]

Delivery signature and integrity

[frosty hemlock]

Ahh

[wide reef]

Oh my bad encryption

Not delivery

[frosty hemlock]

Mhm

Nice

[frosty hemlock]

@wide reef my backup application is becoming a 2 man project now

[wide reef]

ayyyyy

[frosty hemlock]

Im gonna focus more on delivering new features and experiments and my friend is gonna work on refactors

[wide reef]

you go bestie

Would love to use the product some day!

[frosty hemlock]

Yeah, me too, im already using it rn but more in a testing way

And im currently migrating from no framework to vue3

[neon oriole]

it was like a selfhosted backup thing right , im working on something that (what i used that non-uuid thing for) that may verry well hook into that 🙂 its a shell history tool that logs every command but also the environment (the important parts) that that command was run in, who what on what system but seemlessly uss the shell's own history to get what you needed , same user different system, run a commnand on the first system press uparrow on the second and you see the command there also :), i actually had it finished for one system and flat files as db (as good as ) but then i decided maybe multi systges and use sqlite that gets synced to a local server or so

[nocturne galleon]

For the people working with software, is there a way to package a streamlit app using code obfuscation tools like pyarmor etc. that can hide the code sufficiently to discourage people from stealing? Is there a better alternative? Is the free tier pyarmor good enough? Cannot go the route of client-server!

[soft fern]

Anything delivered to the client unencrypted or with keys accessible by the client for decrypting can be poked at or disassembled. Given LLMs/AI can allow folks who have no idea how the code works to vibe code an alternative, unless it's something extremely novel in its implementation you'll need to protect the thing through other mechanisms (licensing, accounts with online login etc). Personally work in the "web space" on lots of client/server things typically secrets are only secret if they are server side only.

[frosty hemlock]

Yeah, ive now for 99% rewritten the whole frontend in Vue3

There's still some bugs in it but most functionality now finally works

[frosty hemlock]

That's also an interesting project

[shrewd valve]

frontend for borg or smth custom?

[frosty hemlock]

borg?

[shrewd valve]

https://www.borgbackup.org/

been around for some time and is getting a fairly big rewrite atm

is purely for backups though

[frosty hemlock]

Mine is custom

Also purely backups atm and written in flask and now also Vue as frontend

[shrewd valve]

been thinking of making a sync+backup server client thingy similar to the drive programs from proton, google, microsoft but not getting to it cause busy with other stuff

[shrewd valve]

oh nice, you got a repo for it or only private for now/ever?

[frosty hemlock]

https://github.com/JordyDevrix/upback

It's severely vibecoded tho, so look at own risk

[shrewd valve]

used to that from uni, dont worry lmao

[frosty hemlock]

Lol

[frosty hemlock]

Ive just finally finished fully rewriting to Vue3

It also looks terrible but its already 10x better than using pure HTML: https://github.com/JordyDevrix/upback/tree/feat/vue

And now im gonna sleep

[frozen flame]

Python code is not hidable from a sufficiently determined poke-abouter

One way or another there must be code (which can be bytecode, but bytecode is easily reversed to regular code) for the interpreter to run

[nocturne galleon]

Noted! Thanks!

[neon oriole]

in some form it already exists , but they host your history and you have to remember the systems-name in order to retrieve any commands , so if you do a reinstall and dont know what you gave the previous sytem for name in the program you have to mail them and they can send you a zip file. So i dug in their code and realized it was a bad setup from front to back. hence i figured make it myself and selfhosted or possibly future with git/hub , did it all in bash first but ran into some gowth issues where if i wanted to add a feature , it became exponentionally more work

so i done it in python now , and using the Click commandline package as only depency so far , that thing is just to good :), once i have something working again ill post he github link here 🙂

[frosty hemlock]

Ah haha nice

[frosty hemlock]

Click is very neat

[neon oriole]

not promoting this but then you know what i was talking about :https://bashhub.com/

[celest ginkgo]

what does bashhub do that regular command 'history' doesn't ?

[rain sparrow]

funny I corrupted yesterday my zsh history XD

[frosty hemlock]

I just got a new idea to add to the app, ✨ AI enshittification ✨

AI generated summaries of your backups

MCP tooling so you can add it to OpenWebUI

I also got several real features I want to add:

• backup run info so you can see who created a backup e.g: cron for cron enabled backups, devrix for a manual backup done by devrix etc...
• multiple server instances feature where you can create a cluster of backups applications on different server and manage them through 1 web UI
• tagging so you can tag your backups with user created tags
• backup removal by cron to automatically remove backups after a specific amount of time has passed
• backup ignore feature where you can exclude files and folders from your backup that are not important
• some kind of notification feature where you can enable notifications if a backup failed
• authentication with RBAC

[neon oriole]

im making it pretty modular where if you want to add a new shell to the thing it should be pretty simple, and your not the first to think ai enshittification, i had in mind optionally to have an ai describe in one line what a command does and store that with the command

but im mostly stuck atmat how to organise the sqlite db's in a sane manner, since commands are added one by one , but depending on the current setting whole db's should be concatenated and or merged so it loads into the shells native history... so when you press up arrow you get the commands you selected for

[frosty hemlock]

If you'd like you can fork it, but we're currently at a very early stage of development so anything can and will change and anything can and will break

[reef tangle]

Anyone here use Matlab?

[wide reef]

Yeah why

[frosty hemlock]

Nope why

[old arrow]

Yeah why

[reef tangle]

I have homework and I wanted to see if someone could help me, am an EE stuck in a CS class

[frozen flame]

Unfortunately there is no help for matlab. Only pain and suffering

[old arrow]

They have good online manual pages

[opaque finch]

I'm an electronics engineer working as web dev lmao and getting paid noicely

[midnight wind]

Matlab isn’t that bad

Has some very nice features I like

[shrewd nova]

Hi, I can help you.

[opaque finch]

Bros please help me with a possible thread starvation problem in .net framework 4.8 version web API project.

Firstly all APIs are async and any service method making HTTP request to 3rd party API or DB calls made using EF are async.
Any method which calls an SP using EDMX file are sync but those are wrapped in Task.Run
Like this

var data = await Task.Run(() => { using (var db = new Connection()) { return db.FindUser(Phone, userName).ToList(); } } ).ConfigureAwait(false);

We can say about 30 APIs are used AND around 10-15 of those APIs use task.run to call SPs.
An SP may not always be as simple as find user, it is required that we use SP in some cases.
Others make 3rd party calls or use EF to make DB query.

This system completely collapses under load. CPU load/utilization is under 50%.
Cant even handle 1000 req within 10 mins.

When I set minimum threads to 250, system performance improves, atleast for 100 user 30 second load test scenario.
I am yet to try setting it to 500 or above.

In this case should I avoid using Task.Run for calling stored procedures ?
What should be my approach ? unfortunately i dont have access to deployment and load testing environments so I have to ask someone from QA team to test and devops team for utilization figures.

[plucky kiln]

Wondering, what do you usually use to secure a AI agents? Twingate? Or some API gateway?

Designing a new system, any kind advice? 🙂

[austere bobcat]

This feels an awful lot like a bot message. Are you a bot?

[shrewd nova]

what?

[austere bobcat]

oh ok. Just felt like a pattern similar to bot messages I’ve seen on Discord in the past. Nvm

[turbid moss]

If you run an LLM locally check this out and let me know if this is going to be useful. https://github.com/Jiss-AI/jiss

[cloud quiver]

Hey, I'm on windows home and I don't have access to the sandbox. Are the free alternatives on the market as good as the native windows sandbox on the others versions?

[hybrid radish]

Hello, me and two friends will work on a project (mostly vibe coded) We chose Asana for progress management. How we can collaborate easier? To combine each other parts and test etc.? I see GitHub and Docker/Podman as options. Anything recommended?

[limpid reef]

Windows sandbox is the simplest, but I prefer VirtualBox because the VMs you create can be kept, nuked, or snapshotted, as required.

[turbid moss]

Just curious, did you try my project yet? I added an easy way to get api tokens without needing email verification.

Would love to hear your guys thoughts

[limpid reef]

I don't run local LLM's, so no, I have not tried it.

[cloud quiver]

Oh ok cool. Thanks!

[hybrid radish]

Anything to this?

[frosty hemlock]

GitHub is your friend

Docker is just for running applications, docker kinda helps solve the "It works on my machine" problem

Also, what exactly do you want to collaborate on?

I've used

• Jira/Tiaga.io for scrum planning and backlog
• GitHub/GitLab for version control
• Confluence/Teams for documentation
• Teams/Discord for communication

[shrewd valve]

GitHub can also be used for planning and backlog via their Projects feature. Currently using that for a Uni Project with multible people

documentation also via their wiki feature but havent used that yet.

look at gitea if you want github but selfhosted

[frozen flame]

and introduces so many other fun problems :D

[frosty hemlock]

Or Forgejo or GitLab which are both also selfhostable

[frosty hemlock]

Yup lmao

[frozen flame]

yeah didn't we excommunicate gitea for going corpo or something

all the cool kids use forgejo now

[frosty hemlock]

I use github

I have my own server but I don't trust myself hosting my own VCS platform

I just know im gonna mess up something and Ill lose all my repo's

[shrewd valve]

havent looked into forgejo but gitlab looked very uninviting for selfhosting when i last looked into it

[frozen flame]

gitlab is... special

[shrewd valve]

still mit licensed with active development. havent heard of the drama myself

[frozen flame]

i think the MIT license was the problem maybe?

it moved from open source to open core

idk i dont care about drama i just heard from the grapevine that we've excommunicated gitea and now we use forgejo

[lean valley]

Framework!

[frosty hemlock]

GitLab is pretty doable but just kind of heavy for smaller servers

I've set it up once on my server and I use it on uni

[frozen flame]

Sir this is a Wendy's

[shrewd valve]

Whats up with the increase of cv like postings in here

[mental panther]

Get your chat gpt ass post outta here

[frosty hemlock]

Had done some small bug fixes now and added a cool background and semi transparency

[shrewd nova]

hello, how are you?

[gilded moon]

https://github.com/pivotal/workstation-setup/issues/87

yooo

(in case this helps someone else)

[spare dove]

how do i manage a git repo now it has people forking it and is semi active

can i continue messing about in main branch or what

[old arrow]

sure, although if you're breaking stuff intermediately then ideally do that on separate branches
Commit or merge to main when a new feature is no longer broken

[spare dove]

ok ill do this thanks

also people said that I need to use cargo for a large scale project is that true?

for me rustc is fine i dont plan on any crates

[keen sorrel]

Are you maintaining a Makefile or something to keep all devs in agreement on all the compile flags? Cargo adds features and profiles to make this easier, not to mention just general configuration settings. Cargo does a fair bit more than just crates
https://doc.rust-lang.org/cargo/reference/features.html

[spare dove]

all devs is just me

and i am using rustc as i only touch standard libary

[keen sorrel]

You said people are forking it. Features are conditional compilation you add to your own project. For example you could gate a C ffi behind the ffi feature. None of the things I've mentioned about cargo have anything to do with whether you only use the standard library

[spare dove]

yeah they fork MY project they can deal with how i do stuff

[keen sorrel]

Lmao okay

[spare dove]

anyway all are in active and just like 8 commits behind

[fading osprey]

Respectfully, have you considered a mental health hospital?
Are you ok?
Who hurt you?

Why is all of your code in a single .rs file
why are you not using cargo?
why are you rawdogging rustc commands to compile this?

There are so many standard ways of doing things which you simply seem to have elected to actively ignore

I have so many questions and every single one of them involves a therapist.

[spare dove]

lol

i just wanted it this way as its simple and easy to read

[fading osprey]

I do not know your pronouns, however, I do know one thing

you, good (gender nonspecific) sir are truly a psychopath

[spare dove]

lol

i am indeed

[fading osprey]

please for the love of all that is holy use the mod syntax at least

[spare dove]

nuh uh

we embrace the mess not clean it

[fading osprey]

I... congrats, you're officially the first case of a developer actively making their own life worse by ignoring the standards despite knowing they're there lmao

[silk eagle]

yuck

[spare dove]

I like it this way

I don't like cargo xd

[fading osprey]

Cargo is great, I'm curious why you don't like it

[spare dove]

I just dont

[fading osprey]

https://tenor.com/view/why-fuck-annoyed-gif-12098885

same energy ^

[old arrow]

You're really supposed to use cargo, even if you're not using dependency crates. If you're looking for minimalism I really think you should just use C

if you have specific things you don't like about it then people might be able to help find workarounds, the advantages usually out weight disadvantages

[neon oriole]

contrary to what you prolly believe i dont hate cargo, but it scares me because it invites the behavior i talked about before, make stuff easy to use and powerfull, and people will put anything on it because its easy, make it hard and nobody uses it so there is no right kind of solution there i think

yea maybe no rolling releases and only release major versions and only one of them such that you always have a finished produc

[merry bolt]

Hello everyone, i am self teaching programming and would appreciate advice. I’m unsure which project to build that would help me learn fundamental concepts, instead of spending most of my time on APIs and framework-specific mechanisms such as signals and slots.

[fading osprey]

Well, what programming language are you learning? This will affect what you can reasonably build in that language, and will help people suggest better resources to help you! 🙂

[cloud knot]

today i wasted like 3 hours with some Android fragment not showing up (well, black screen was showing up), just to find out the configuration for said fragment had a "showPreview" boolean switch which for some reason allows you to disable UI of a component which main feature is to show the preview of the camera... and if you set it to false, then it shows just a black screen. 😄

[merry bolt]

C++

[proven coyote]

Been working on some Nostlagia bait for an Android launcher

[spare dove]

submitted my first pr for a project ive been watching for a while

feels nice

[simple canopy]

Very nice. It does feel good

[spare dove]

got declined brutally and told to re do it

[simple canopy]

RIP

[spare dove]

it isnt that hard to fix

i just need to do each fix in a different pr

[cloud knot]

ah you sent the noob mistake of giga pr solving 20 issues in one pr. Normal reaction tbh

[spare dove]

3 mistakes fixed

[cloud knot]

yeah, single PR per single issue is the norm everywhere

[spare dove]

Never worked outside of my projects before

and in my projects we just flip it and push always works

[silk eagle]

thoughts on https://github.com/dotnet/aspire / https://aspire.dev?

[spare dove]

https://github.com/akirathelinuxguy/Shelly-ALPM/pull/7

im so proud of myself fully merged and all working

[wraith turtle]

hoping someone else might enjoy the hilarity this is.
to clarify this is not a remote display or remote session, it is a webos called puter but i forked it slightly and now am running a "linux subsystem" so linux subsytem for puter effectively with a debian base. yeah it only supports 32bit but.... its proper debian but kernel-less. im using cheerpx which is a translation layer that converts x86 to wasm so its running it natively in the browser 😅

[frosty hemlock]

wth

Can it run doom?

[spare dove]

src?

[wraith turtle]

im in the process of creating a wsl equivalent for the puter webos using cheerpx

[wraith turtle]

once i get the display outputs taken care of to get graphical output in the desktop yes

[spare dove]

would it work on linux host

[wraith turtle]

its not quite how it works, its an os in a web browser and im adding a linux subsystem to it so you can run linux applications on it directly in the web being translated to wasm

you can host it on a linux server though, yes, but everything in that view is running directly in the browser

[spare dove]

thats nice

[wraith turtle]

its not emulating or a remote desktop or streaming, its running puter os in the web with a linux subsystem that can fully run 32bit linux applications with a debian base, you can even apt get packages and install them from the debian repos

well, once i finish getting networking working lol, you can run the command but it just fails because of a lack of a network

like its got a full ext2 filesystem

[spare dove]

whats a good license again

which forces forks to stay open source and everything open

[shrewd valve]

gpl v2 or v3 and later. depends on if you want to be linux kernel compatibel or not

agpl if you dont want the application or whatever to be monetisable as a service online.

[spare dove]

i just went gplv3

https://github.com/akirathelinuxguy/rtimer/tree/main
readme looks like shite

no wiki no nothing but ill get round to it one day

[neon oriole]

depending on the confidence i have in the mess i publish, i either go with Unlicence, MIT,GPL , or All Rights reserved (as in no licence yet)

[spare dove]

i went gplv3 for the main branch and all the forks are using gplv3 and no one is complaining

[neon oriole]

tbh im actually in favor of a whole new category of licences... a licence requiring anybody using the code in another project (be it open or closed source ) must inform the licence holder of this, and this is valid for subsequent uses of other projects.
ofc some sort of system should be put in place making it easy to comply by running a git command or whatever that mails all authors in the chain that you are using it , this could be restricted to only if you use it and redistribute your code...
but of some of the tings i made that i forgot i shared , damn they have users,... , and for others that i think should altleast have a couple users... but it couls also help determening wich projects are dead in the water and wich ones deserve more time and attention

[spare dove]

id really love this

you could create your own license for it

[winter jetty]

https://codereview.stackexchange.com/questions/301160/encrypted-messaging-protocol-over-ftp-in-python/301162#301162
tell me your opinion about it there pls

[spare dove]

post it to a repo

[shrewd valve]

repo is at the bottom of the post

gonna be honest id remove that phone number from the readme though. thats never a good idea

also it says in the readme the project is licensed under creative commons but the license file is the mit license

[unique wigeon]

Idk if this is the channel to post this but I’m gonna learn c++ the hard way by doing a project with 2 of my mates

[winter jetty]

Good project💯, keep going

[proven coyote]

Back again with more search integration and inline reply on search

[unique wigeon]

idk how long its gonna take

multiple months minimum probs

[unique wigeon]

giving up and using unity

(watching initial d if your wondering)

[silk eagle]

chatgpt will randomly give people your phone number

and its now permanently on a spam call list

[lean valley]

Progress being made???

[raw musk]

What if you use google voice or something like that as a 2nd number, is it fine to put your google voice number on a readme?

[keen jungle]

@silk eagle yes but how to stop these so many ads

[unique wigeon]

Fuck it we ball

[neon oriole]

trying to do it the right way ,... meaning ... trying to use python asyncio ... it breaks my brain really :S going back to using process signalling and inotify , seems less of a hassle unless i can find an explanation for how asyncio works (the ones explaining how to work with it dont seem to cut it)

[unique wigeon]

I’m using c for my engine

[old arrow]

I wish I had a good explanation of async and coroutines, I only learned by doing
But if this helps, a coroutine is basically taking your function and turning it into a state machine, where each of the states is one of the pause points (when you call await). The function can be paused at these points to let other coroutines run on the same thread; this is useful because there's often things (like I/O) where you are just waiting and can progress without any CPU time. Then there's some notification or timer that's used to keep track of when the coroutine should be considered ready again, so it can be queued to run on the thread again

Take for example a network request, you might do some work to send off a bunch of packets at first, but then while waiting for a response to come back (due to network latency), you don't have any useful work to do. So you can await: give control of the thread back to the runtime, and give it a signal handler that'll tell the runtime when the network request has some response data. Then the runtime waits until it gets that signal before executing you back on the thread

An async runtime and often does use epoll(7) to handle waiting on all these notifiers for all the coroutines simultaneously. So you don't have to spin loop on the CPU in userspace constantly checking or anything like that

[neon oriole]

well i more or less got that already , and i can construct the basic examples that you can find in every tutorial, but im missing somthing in terms of background i suppose (how it actually does operate in the background) what ecactly is an eventloop how does it work , can i have more than one eventloop , if not why not ... if not can i have more if i use multithreading , if asyncio.run() creates an eventloop and runs it , whats whith loop asyncio.create_event_loop();loop.runforever() eg or if i create a task , but the actual function is a clojure that uses properties of the object its a method of ,...
eg

[lean valley]

Laptop 16 fingerprint scanner is working!!!

[idle shadow]

For all the gods out there. What would be best to learn if you want to work as a fullstack web developer? Python? Angular? JS? Typescript?

Not familiar with typescript too much. I don't know how it compares to javascript. (I know it compiles to javascript code)

Ive been using Wordpress as CMS for now, where i guess php would be the better option at that point, but thinking of moving to headless CMS and eventually work with Swift

[neon oriole]

I would do anything to avoid javascript based stuff , but ... as a fullstack webdev well id still go for python+php+svg+css+html with as little js as possible (most of the stuff js is used for can be done with svg and css anyway)

[idle shadow]

svg? is that much more than just convert shit to? Or are you talking actual graphics design?

[neon oriole]

but you prolly have to know js anyway because its around so much

[neon oriole]

you can style svg , with js , and you can animate it , so your fancy js button , could just be a svg with some css effectys

[idle shadow]

Well ive worked with html5+css+js for some time so its not unfamiliar at all. Node.js, angular.js etc is barely touched on, and wouldn't know a thing. I can't write code without looking up, but i do understand the structure and how it works for the most time

[idle shadow]

Ah i see. I've always been interesting in programming from young age. Started learning Java at age 13, then tried some C# and C++, then standard web language/markup

However i was able to produce a few kids and never got to go get my bachelor/masters lol. With the mother being an ex at this point i wanna commit again 😄

[neon oriole]

advantage of js is that you instantly see results of changing smth visually represented in the browser, but its a world of more pain and headaches (your taking on depencies like mad you have to stay on top of , starting a project gives you 120k lines of famework not even related to your project) , on the otherhand learning python is prolly keeping you on the terminal for the fist year , before even attemting smth like a webpage. django is popular but if you do want to get smth lighter simpler in python, there is pylons/pyramid i think

https://docs.pylonsproject.org/projects/pyramid/en/latest/

[idle shadow]

I hear python all the time.. Also heard about django yeah. I guess python is where i should go. I guess PHP is also quite common still. From what i can tell, powers 75% of all websites? I guess thats because most major CMS use php though.

[neon oriole]

php never left idd, and its gaining popularity again, prolly also because its origins are basicly to produce html pages as output

[idle shadow]

And are Node.js and Ruby used much? Javascript, C# and Java are very easy for me to read. Gonna be interesting with Python

[neon oriole]

pythons less typing thats for sure 🙂

[idle shadow]

kk i guess ill focus mainly on learning Python and php, as my html5,css,js knowledge is little better than beginner

[idle shadow]

Thats cool 🙂 So Python with Django framework?

[neon oriole]

there used to be a verry neat thing for setting up a local development server but i think most of it is payed now , ampps

[idle shadow]

Oh yeah thats fine. I run a few websites already 🙂 I'll increase my personal one and make a new webserver

[neon oriole]

well if its smth, but like django is big , and maybe to massive to start with , thats why i mentioned pylons and pyramid , wich is basicly the same concept (but different) but in ultralight mode 🙂 and it comes with its own html server to run locally

[neon oriole]

yeah i wouldnt do dev on a prod server 🙂

[idle shadow]

One thing though. People rarely do web development from the bottom nowadays (right?). I've heard headless CMS is the way to go. Anything you would recommend me getting into? Where i can intergrate my upcoming Python knowledge? 🙂

[neon oriole]

just smth you spin up localy on your own system to test stuff 🙂 there is ampps , xamp , if on linux , you can just lampp , pylons comes with one

no idea, the only cms ive used and liked is this one (and its pretty unique a that amongs cms's : https://modx.com/)

because there is no formal strucure imposed on anything with modx , but that also means not much of startup boilerplate pre configured and ready to go eg

and modx is php so , but you could combine it with python aswell 🙂

[idle shadow]

Most familiar with LAMP from before but i've come to the conclusion i need to adapt myself so using nginx only now 🙂 Ill look at both a stack with php and python and perhaps modx 🙂

Thanks a lot for the help 🙂

[neon oriole]

think i have that asyncio almost figured out now....

[neon oriole]

think i got it 🙂

and its done with every log represented as an object and taking care of its own monitoring for changes ,and a recursive watch that watches for new logfiles , and creates the object for it , after wich the object also starts monitoring itself ... ,what still doesnt work is when i delete a logfile 🙂 fd filedescriptor errors ... 🙂

[verbal lily]

Anyone here familiar with Python?
I don't know if i'm having a skill issue, or there's something up with my quiz (Practice quiz, infinite attempts before you say)

[limpid reef]

pretty sure you need to do some more mathing in the following section: 😉

if partial_block_remainder > 0:
    return 4096
return 4096

without extra mathing, this will always return "4096" no matter what number you put into calculate_storage()

[verbal lily]

Oh whoops, didn't post the question
Lemme find it again

[limpid reef]

you did post the question... it's in your screenshot?

[verbal lily]

Oh yeah lol
I panicked for a sec, mb 🤣

[verbal lily]

Maybe there's not supposed to be 2 returns?
It will be 4096 for both, since there's no else(), right?
Or elif()

[limpid reef]

Neither of the return lines should be returning a static number. They should be calculating things. 😉

[verbal lily]

https://tenor.com/view/james-franco-wait-what-wut-wtf-nani-gif-25479769

Huh.... OK

[verbal lily]

Can you point me in the right direction, without just telling me what it is?
Even though this is pain, I still want to figure it out lol

[limpid reef]

Sure, you need to calculate the total number of bytes a file would occupy, rounded to the closest full block size that can "hold" the file. In other worse, multiple your block size by <something> then add one block size to it.

I don't know how else to explain it without giving away the answer so yeah.

[wraith turtle]

so i got a little bored and was playing with rust. got a basic electron like system using the servo engine and nodejs instead of chromium

currently implementing ipc but it can properly load an app and has a nodejs backend but i still need to get it communicating bidirectional

[old arrow]

nice; how much work did you have to do to make this work

last time I messed with servo it was a bear to even get it to compile on anything other than a very specific rustc version