#programmers-off-topic
1 messages · Page 17 of 1
Luckily I've only had to read Java (to port an Android app to iOS), but man was it verbose and annoying to read
C#'s types are also pretty standard so their use ports to other languages pretty easily
I agree with c# being a good choice for a first language
It is fairly high level, with a GC, and a well developed ecosystem
I have super strong feelings about Python/JS being recommended as a first language.
JS is like the opposite of that. Lots of weirdness to have to account for, no standard IDE with something like Intellisense to help beginners, no compiler with useful errors, no strong typing
Tbh I feel the same about python
The first ide people tend to start with - IDLE - is trash
Vscode is good though, as is pycharm
While you can type your python and you should many people don't
i will never develop character. python can stay in hell where it belongs /lh
I don’t know any JS but I’m still wildly against it as a first language
I will begrudgingly concede that it’s clearly useful to some people but it’s such a weirdly structured language that I expect people to take away some very weird lessons about programming if they only know JS
I think it depends what the language is used for as well. I agree it should be a statically typed language, but Im more lenient if theyre using some visual or game engine to help convey concepts. Then I think you can used dynamic stuff more
I agree that if someone has a clear goal in mind my stances on which language is best can definitely change. I also think having a goal can be very helpful when learning, as it gives direction and motivation.
i wouldn't recommend javascript as a first language either.
it has some very good things that i like (expressiveness is quite high, map/filter/reduce, etc.), but it also has a lot of wat.mp4 and poor choices from the 10 days that we are still stuck with. basically the newer parts are pretty good, lol
We used Processing quite a bit when I first got started, which was in Java, its not a bad way to go
(also take "newer" from me there with salt. i haven't done javascript in any real sense for a few years so it's probably got a bunch of new stuff again that i don't know about)
just pick a language and stick with it unless it's bash then pick something else
new stuff means new jank!
I'd did js first and turned out fine
My first language was processing js for the hour of code haha but I'd probably recommend python even though it's weird
A lot of courses start off in Java
My favorite ice cream is graham central ice cream because it has graham crackers and it reminds me of a train station in New York City grand central station. I have to take lactose pills to eat it for toppings m&m whip cream and hot fudge sauce and rainbow sprinkles
uh
I did python first and did not end up fine
I did BASIC first
I feel like doing Scheme first didn’t have much impact on me 😛
I can just say, when you will start learn to code then start without IDE and Intellisense, because you will never learn the language properly. Also did this "helpers" often do a lot of garbage. When I learned coding did IDE's not exist and the same on Intellisense. My first language was Basic & Z80 Assembly for Amiga. At this time we hadnt online repositories like Git or Turtoise. We had to sync and manage the project by ourselfs. Also the code which was written at this time hasnt any framework or something like that. We had to write our stuff like that on our own. Was a different time, but the programms which was written was 100 times more efficient.
Writing a game for my Amiga is a genuine bucket list item.
Programming on Amiga is funny and its like a bit you would work with Batch with a good amount of more power.
I just wish I could think of a pun like the lore behind Guybrush.
That would make it perfect.
i should dust off the super famicom 65816 assembly and make something one of these years
some of those old consoles have pretty robust toolchains these days
Yep, for GB I can recommend to use GBStudio. Its like working with Unity just for GB. Also not much Z80 Assembly knowledge needed. As I know is the compiled ROM usable on real hardware.
Really cool. I already tried it, but the Famicom is a bit special in the Instruction Set. For example, some special instructions like HDMA is a bit complicated to use.
I feel like for me, half the fun of making games for older systems is not having access to all of the modern tools.
to a point. compiling to some of those system can be a royal pain
This has a good reason why compiling for those old systems is pain, because we havnt a Counter System for that oldies like this one: https://upload.wikimedia.org/wikipedia/commons/thumb/3/34/NES_Test_Station_%26_SNES_Counter_Tester_20160814.jpg/1280px-NES_Test_Station_%26_SNES_Counter_Tester_20160814.jpg
i made a really shitty pong game for game boy years ago
Its a lot easier when you have a counter system to get H-Blanks and other timing related stuff which is otherwise not accessibly through software.
i know the tools have gotten much nicer since then
is the H blank value not stored in a register somewhere
Nice. I just tried if Im able to show a picture on the screen. Then I dont get further.
No, because H-Blank is time discrete and can be calculated. On PAL 50 Pics/s and NTSC 60 Pics/s. Some people will not understand that Frames per Second is not equal to Frame Rate.
But Im also not a professional at this point. Its just what I listend and learned over the years about that topic.
I've written game boy emulators in the past, but thats the extent of my knowledge
Thats impressive. Respect for this. 👍
If someone would learn a bit about that topic in a good understandable way. Go to Youtube and search for "Retro Mechanics Explained". The videos are pure gold and easy to understand. Also did Retro Mechanics Explained videos about really exotic gaming systems as the ATARI QuadraScan.
That's a great channel
I wanted to make a tempest emulator after seeing their video, but I have no idea where they got their documentation, I couldnt find anything
And i dont exactly have the means to reverse engineer a tempest cabinet lol
Most old arcades using Zilog Z80 processors. You can get good sheets for the instruction set online.
(Fwiw, I don't buy that intellisense makes you a bad programmer)
Tempest 6502 I think
I guess I meant more for how quadrascan instructions were encoded and such
people say this?
intellesense doesnt make you bad, but I've definitely seen teachers rely too much on it. I had a course in Java that was more like fill in the blanks than actually learning how to program. Honestly, most of my Java schooling was like that
ah ok
Sorry for the missunderstanding and my sometimes broken english or not able to find the right words. Im a trained english reader, but my sentence building abilities are crap. XD
What I mean is what aquova said. Also did I experienced when you take Intellisense and the IDE away are a lot of people not able to programm anymore, because they never learned the language properly. They just learned how to use the IDE and without the IDE they can do nothing. Programmers just needs a computer, a text editor and an interpreter/compiler. That other stuff is just to speed up the workflow.
huh, i just learned that AGI doesnt exist yet? or didnt as of 2023 idk about now though
(Artificial General Intelligence)
wonder what that'll be like when its made
it wont be made lol
Lmao. Yeah, AGI is decades away, maybe centuries, assuming it can be made at all.
im just now finding this out LOL, i thought cluster AIs where AGI
but ive found out they are different(dont know the offical name)
oh yeah i fully know that
i thought that AGI was like our AI but just "bigger and more"
learning something new everyday!
AGI is what theyre calling real AI
Ahhhh ok
because theyve overloaded the term AI already lol
they really have :(
The main reason AGI is so difficult is that it requires a complete understanding of the physical mechanics of abstract reasoning, which we don't have, and a practical way of converting that to a reproducible hardware design
like what is AI even now :/
everyone just throws the term around like nothing
Interesting!!!
I hate LLMs with a burning passion lmao
i wonder if they will be made once quantom computers come into a thing
lots of people do
Umm... AI stands for Apple Intelligence /s
I mean, AGI is a term that's been around for a long time. It's a specific subset of ai that is genealized instead of narrow. IE it can do many things not just one specific thing
but they dont deserve the hate they give, just they have gotten a bad rep
(at least from what ive seen)
Apple Intelligence is so smart you can now move icons from the of your home screen
:O
I've been asked more than once if I'm using chatgpt to write my dissertation and I was like. I would rather launch myself, and also you, directly into the sun
im down for friendly banter but dont mean to upset anyone btw
Do you actually know why they have a bad rep?
the hypothetical, general "you", not directed at anyone here specifically
honestly no, everything ive seen is just people saying they suck or are stupid. not a single actual responce :(
There are legitimate ethical reasons
welp, now that i'm back i should probably vacate this channel for several minutes
Alright, I'll explain it. Just gimme a sec to get on desktop
<3
I get why lots of folks have the gut reaction that "AI" is cool. cool =/= useful, accurate, reliable, ethical, etc
yeah, ive had a few moments where a friend coded some stuff and people here thought it was chatgpt... nah just them using old techinques in there code 
thats what i see people do soooo often, like they think ai is law in its words... NOOOOOOO
oh?
Sorry :(
(We're about to get the Wren/Quill combo explanation I think!
)
hey the more info the better lol
(I can't exactly contribute from a programmer pov, but hoo do I have things to say as a lawyer)
I say say 'em! It's relevant to the topic.
knowledge!
i like learning lol, even more so when it helps others(or helps others through me not being stupid
)
my work is in public health, i will lurk and add perspective if needed
how convenient there's such a wide range of professionals in here rn
from a legal perspective, the likes of ChatGPT and other LLMs are extremely problematic from a copyright and privacy perspective, which ties into the cool =/= ethical etc comment made by magitek earlier
i opened a port on my PC and dear lord the amount of IPs that have tried to access it is astounding. thank god for whitelists 
Ah I see
I think intellisense mostly stops me from making stupid mistakes
(Quick what is python for make directory?)
And even ignoring all ethical issues, the primary concern for the person using generative transformers (not calling 'em AI, screw perpetuating that) is that (at least in the US), anything produced by such technology is inherently not copyrightable because it wasn't created by a human.
the element of consent is absent in scraping, because very few companies actually ask before scraping work available on the internet, or they try to force you to agree to said scraping - which, in the absence of actual legislation... is not good.
and yes, anything created by Gen transformers isn't copyrightable at all
Often I know what I want to do already it is more the last mile of "and what was this function named?"
there's no human element anywhere, which is a necessary element to copyright something in the first place
(this is more or less universal - most countries are signatories to the assorted IP conventions, which standardises things somewhat.)
Which would technically still apply if you trained your own model 100% from scratch exclusively on content you own.
what if a human hand wrote all data for the generation? would it then be copyrightable?
so, to start off, I want to say that in technical terms, LLMs/neural nets are not a new idea- the idea has been around for decades, it's just that up until this point, we weren't capable of producing hardware that could actually running it. (partially.) Conceptually, it's a simple idea, and there's nothing ethically wrong with it- the basic way that it works is that "correct" output data is used to create a kind of mathematical prediction matrix, so when you feed an unknown input, the prediction matrix calculates the most likely matching output.
In practice, unfortunately, there are ethical problems with all of the LLMs that exist. Because LLMs are essentially brute-forcing, you need a huge amount of training data to get any kind of even remotely accurate results, and that doesn't come from nowhere. all image llms are based on literal decades of online sweatshop work labeling images. Of course, most old-school captchas were also helping with that, but it was only a small part of og a MUCH bigger project. The use of so much underpaid labor is a problem, but unfortunately, while the original image set was broadly licensed/free use images, later data sets, in particular for artistic LLMs, came from scraping petabytes of unlicensed art and photography from... basically every website.
This usage of unlicensed content without permission, unfortunately, extends to EVERY LLM currently on the market. All of them. Even the ones that claim to be ethical. (cont...)
i wouldnt copyright my stuff if AI made it any ways, doesnt seem right to
cause i didnt do the work, the machine did
according to us case law, generated content cannot be copyrighted anyways
This scenario has not come up in court yet, so it remains to be seen if legislators anywhere will fill that gap
hypothetically, no.
because the human element was the creation of the software, there
the actual software doing stuff is not considered "human intervention", because entering a prompt is considered negligible as a catalyst for creation
I can't say anything about college programming programs, I be the E in ECE
interesting! i can see how that would be confusing to some people in court
(Despite being a better programmer than a lot of people taking a bachelors in CS!)
to be clear - this view was only formed after experts were consulted. amicus curiae are very useful!
oooo, lots of info! i didnt know that all llm's were "scappers"(as i call em)
new thing learned, amicus curiae. fancy word lol
The more high quality data they're fed, the better they are.
it's latin for friend of the court!
we use a lot of latin
Can you imagine how crazy it's gonna be when computers can operate at the same scale/complexity as human neurons, and there is a debate over whether AI is sentient?
sometimes I think code is easier to understand
an additional problem that's happening now with the art/conversation-focused LLMs is that they've metastasized and are now feeding each other garbage data, producing increasingly garbage results
@supple mountain You leveled up to Cowpoke. You can now speak in our voice channels and share images in all channels!
thanks, gov'na
Did you see the pcb layout llm?
Just to highlight just how much data they're trained on.
true, thats kinda why i wanna make my own AI chat bot using my own data... but that proved to be... not workable at the scale i had it 
Tbh I can see the value of llms sometimes for pcb layout- trying a bunch of shit and seeing what works is great
yeahhhhh, found that out realllll fast
Making my power and signal traces the same size though
more amusingly, the garbage is a consequence of unrestricted scraping. glaze and nightshade work wonders for fucking up generated artwork
yeah i do think they can be useful as a brainstorming tool, or a framework for throwing shit at the wall to see what sticks
Oh, this reminds me of something that might entertain you. I'll DM it!
I wonder how much can AI improve at this point, it's already been trained on the majority of useful data that exists, and new (real) data is not being created at the pace necessary to outrun garbage data being created by AI
but from an ethical (and legal) standpoint they're untenable simply because they're plagiarism machines
What I'm waiting to see is the effectiveness of the new AI regs in the EU - and whether or not other jurisdictions will catch up.
it can improve a lot... provided its uncensored cause theoretically you could just kinda.... feed it the entire internet... but ethics there are not good
my understanding (y'all would know more about this than me) is that the way buzzword AI works under the hood is too restrictive to do what people seem to think it does
The problem is it's already consumed a significant portion of the useful corners of the internet
on a fundamental level, LLMs simply pull shit out of their asses that looks right
It just a big ol' system to predict what's most likely to follow the last thing.
the ethics there are flat out non-existent, yeah
So as it continues scraping sites like reddit, it's getting content which was produced by itself
Garbage in, garbage out
idk maybe the fact that every chatbot has to have anti-nazi baby rails put around it is a bad sign 🤷♀️
What's worse is that the data they are fed is indiscriminate- meaning it includes things that aren't correct, and things that may appear to be correct, but are actually biased. For example, AIs tasked with determining whether or not someone is likely to commit a crime consistently rated black people at higher likelihood than white people. In a sense, it's just bias/misinformation automation unless you're very careful about what you're feeding it.
The other problem is not so much ethical as it is a flaw with how LLMs are used. I mentioned before that LLMs are essentially just overgrown prediction engines, which is fine for some applications, like autogenerating captions for video, removing video backgrounds, real-time machine translation, etc. Those things are fine with approximations- if the prediction is wrong, it's no big deal. Unfortunately, LLMs are being used for more than just that- they're being used as a source of "information", and for making important decisions. But they don't "know" anything, they only make predictions, so they often hallucinate lies (bad) and predict what someone else would tell you to do, regardless of whether or not it's actually good advice (worse). It's easy to humanize them because their prediction capibilities allow them to output natural language, but it's extremely important to understand that they don't understand what you are saying or even what they are saying. And it's somewhat concerning when they're being used for things like "driving cars" and "military/police applications".
I feel like the technology has potentially peaked for now
Improvements in it are going to vastly slow down for now. Almost guaranteed.
Yeah
I used to read articles on Medium, and I've noticed it's become trash lately
Like I can tell that the majority of content being produced on Medium is AI generated, it's obvious
The improvements will come from filtering the training data to exclude incorrect information, which is... going to be a ridiculously tedious task.
Imo the best use of AI for me is "take this paragraph i wrote and make it sound professional"
"And not like I'm two inches from losing my shit"
i knew a bit of that, like the fact that they basically dont know what the heck is going on. but the rest is all new... and that crime rate thing :(
i do the same but with mediaval stuff, "take this and make it old english and mediaval sounding"
yeah again like, training set stuff aside, llms are fine to use for low-stakes things where approximations are fine, but the rate at which it's just being blindly shoved into everything for investor money is... somewhat concerning
cause fancy words be fun
yes, and:
-
the very idea that we should let the plagiarism machine make life-or-death, highly nuanced decisions for us is dangerous and stupid
-
the very idea that we should let the plagiarism machine do the jobs of highly skilled professionals, but worse, is silly and inefficient at best... and dangerous and stupid at worst
-
the "AI" tools that are actually really helpful are, like you said, very specific algorithms that do very specific, high-volume, low-risk things
to add on to the point of how the data fed to LLMs is indiscriminate, and magitek may be able to confirm what I'm saying - sensitive medical information, private information that should not under any circumstances be used this way, is being scraped, and I cannot for the life of me understand what good someone's medical records are for any use cases of LLMs.
YES.
DEAR GOD KEEP IT AWAY FROM MEDICINE AND PUBLIC HEALTH
Keep it away from anything protected by personal data protection statutes, tbh
What I want to know is how in the hell that data was scrapable in the first place...
AI should not be allowed medical info... google is already bad enough
LLMs are probably not the right application for medical data, but there are good data science applications for medical records.
Sold, perhaps.
money does buy a lot
I hope that we get off the LLM bandwagon and invest in other interesting areas of Data Science/Machine Learning
looks to windows recall in the future
oh for sure, but in the context of LLMs specifically - no.
and even in the context of data science application, you'd need the consent of every patient
(i came back)
using LLMs to make important decisions is bias laundering. whatever bias is baked into the training data becomes bias in the output, which output appears (without further scrutiny) to be neutral because a computer produced it
I think the topic's... mostly died down at this point?
yeah I'm glad that can't run on my computer
oh god I have to be so careful when I buy my next laptop
i would invest in anything that makes my chat bots better and has a more capable "feel"
absolutely floored me that it was originally going to be on by default
Like, LLMs for interpreting language/instructions from conversation/text, yes, but it needs to be paired with special-purpose secondary applications to do anything useful
Recall is an idea I absolutely love, that I want Microsoft or any company to have absolutely NO say in its creation or implementation.
exactly
Recall scares me
If it's 100% open source and I can be sure it's all local and sescured properly, fantastic!
recall caused my bp to spike, which isn't good post-surgery
(it probably can(if you are on Windows), you dont need the chip thingy for it to run. You just need the Dev version)
so I've forced myself to stop looking into it
the case with medical records is a little different because, like, retrospective studies exist. once the data is de-identified, blinded, etc., machine learning to wrangle your data is fine
there already is one, its called OpenRecall
I actually saw that!
I don't think I'd have much use for it but it would still be a good tool... if it were local and not sending screenshots of my desktop to the cloud
though it probably shouldn't be an LLM doing the wrangling because you want the results to be replicable lol
yes, once anonymisation occurs, it's mostly all good
i have it and its creepy what it can do! all in 1 python script as well 
It's not sending them to the cloud. It's just storing them locally in an unencrypted database! 
EVEN WORSE
That's much... better...?
oh yeah that's so much better /s
Re: ai and medical data
I've been using LLM tools as a aid to programming for awhile now, and I feel like it's gotten worse at its job over time
WHO DESIGNED IT
A TEENAGER ON BABY'S FIRST SQL PROJECT?
It is useful in being able to find possible correlations
But they need to be checked again in a proper study
[hisses]
I really think the thing is in a downward trend, and the tech industry is heading into another bubble
Huh
I've never used an AI assistant
Like the same situations I ask it for help with are producing worse results
More AI-generated code being scraped and used to train it!
i would if they were better
if you're doing exploratory analysis, then having something dredge up the most interesting stuff in your data is fine and encouraged (though imo it's far better for everyone if it's done by you, with skills you are practicing)
I was around for the dotcom burst, albeit a child, but I hope we're not heading into that territory when it comes to "AI"
I just don't see the point of using LLMs for programming tbh. I rarely have trouble designing solutions to software problems, and when I do it's almost always something like "how does this library with zero documentation actually work, functionally" or "where's the error in my transpiler" and copilot or w/e is... not gonna help me with either of those things
for confirmatory analysis you absolutely should not be looking for "possible correlations", you should have hypotheses ahead of time and test them appropriately
Oh god
I mean for me personally. I've seen other people use it as a learning aid
I always enjoy having to decompile a library to understand it
I wouldn't say ChatGPT has programmed for me, but I've asked it for ideas/approaches that helped me research solutions more quickly than just a blind Google search
hey, at least harmony and avalonia are both open source
this is what i do, i look at code that works and reverse engineer it, just need the thing to NOT die on me with its code
like ive learned a good bit from messing about with AI outputs and reverse engineering the hiccups to work
hilariously, part of the problem here is that google has sabotaged its flagship product's functonality so much that you have to ask a chat bot to help you massage google into giving you anything useful
Early on, it did seem to help me to get solutions faster, but now I think I spend more time refining the questioning to get any useful suggestion
I'm like, "No that's not what I meant, and I already said that earlier"
yeah I gave up on google ages ago. DuckDuckGO isn't perfect but it's... better than google and bing
i think some of it comes down to how you use google
google only ever gives me horrible slop unless I'm looking for minecraft mods on curseforge
Google rarely has the most useful result on the first page for me. SEO has ruined it.
bog standard google sucks, google scholar is chef's kiss
(Granted, well, half the time I'm googling things like "sn65lvds048a")
carried me through my thesis fr
Which
google scholar
Well. Has one valid search result
is good shit
i am technically gen z and have adopted the bad habit of just typing "keyword verb question" instead of treating it like pubmed or whatever
(what is)
(TI part)
part of it is that I got used to using google's special search syntax stuff years ago and now the only thing that seems to still work is site:
I've actually been using Kagi lately
I'm gen z too, had to unlearn that habit
maybe if you use the google search bar special syntax it's still fine
I've tried using double quotes like I used to but it just ignores them now
(ah, ty)
I've become a bit of a privacy nut lately, trying to get off of services where I'm the product
also all the results are just full of SEO slop unless you're searching something super common. (and even then...)
anyway, i gotta go do some real work, but i think one of the fr most insidious parts of the AI craze is that people want to use it to bypass learning skills and information
Linux is your best bet soon then if you havent moved
Or something super obscure
there is no substitute for the human brain when it comes to nuanced thought. agi is science fiction
this search is not super useful but it is a really great way to discover buried pages
https://search.marginalia.nu/
search.marginalia.nu is a small independent do-it-yourself search engine for surprising but content-rich websites that never ask you to accept cookies or subscribe to newsletters. The goal is to bring you the sort of grass fed, free range HTML your grandma used to write.
grass fed, free range HTML
these are my people
wrt to Linux... I have many questions and almost all of them are for the purpose of convincing my folks that my next laptop really doesn't need to be a windows machine 
machine learning tools can be great for learning aids, especially for folks who are on their own, but the goal needs to be critically thinking and learning to do the thing yourself
YEAH it's great. basically all the results are old-fashioned 90s-looking webpages
oh, you mean "good websites"! i love those!
:O maybe i can find some rainmeter skins here
cause, i cant find em anywhere
The first 500 people to use my link will get a 1 month free trial of Skillshare https://skl.sh/emmainthemoment01241
Mike and Dave bought the domain for knitting . com with a big announcement on their podcast. The knitting community was not pleased (to say the least). Here's what happened.
vest I'm wearing is a modified version of Martin S...
deviantart of all places has many
(Arch, by the way.)
EMMA MY BELOVED
YOINK
deviantart is where I got all of mine. there is so much stuff for customization on deviantart if you look in the right categories
rainmeter, that's a name I haven't heard in a long time
i want an SAO theme
but they dont exist
id litteraly PAY SOMEONE to make one but i aint got the money yet
Anyone remember Konfabulator?
Rainmeter is incredibly easy to make stuff for. there's probably guides on the forums
that's what I told them but "don't be stupid" was my dad's response lmao
he thinks there are too many compat problems
I suppose he would know more than I do but surely it can't be that bad
Did you see my sweater?
before i deleted twitter, i saw a thread by a phd adjunct somewhere and his whole thing was INSISTING that chatgpt was going to replace statisticians. wow, look at the chat bot making a pivot table for me! wow, watch me click a button to make the computer generate a model (something many statistical software packages already do for the CLI-averse)! wow, watch me ask openai to write the background and discussion sections of my manuscript!
the pivot table thing legit had me cackling
Yeah, no, don't use Arch for a day-to-day system unless you're okay with constantly keeping on top of changes you need to account for every time you do an update.
side note but the number of people who are terrified of CLI for no reason is just sad
you have a whole doctorate and you're so brain-dead you can't make a PIVOT TABLE?
i use ai for css sometimes... magic that stuff is(or HTML coloring)
I don't believe so! 
... oh balls.
what do you suggest then 
statistics anxiety is a legit thing and for folks who are coming in from less STEM-y fields (i.e., sociology), the software packages that use dialogue boxes instead of a CLI are less intimidating
css! is! not! hard!
you just have to actually bother to learn it instead of going "oh yeah I know the syntax I can write it"
you also need to learn specificity and the box model, and keep in mind dom hierarchy
i wanna do a Linux Distro built purely for coding envs
I love the pattern you've got going!
as for the "have chatgpt write the discussion section" thing, statisticians are generally not known for being great communicators
This is my concern, tech companies are laying off employees and making heavy investments in AI. The senior leadership has no clue what it is they're getting into. I hope things flip around and suddenly there is a high demand for human labor again before companies go bankrupt.
hard to be math wizard and word wizard at the same time
but like, the AI isn't going to present the paper for you, bud
you have to work on that skillset
see thats the thing, i understood none of that
. Its a language i gotta learn better
like python, exept python sucks still in my mind
I feel like, from what I've seen, that a lot of that garbage thinking comes from the top leadership not being engineering in the first place
The CEO of my company called the people in IT the "computer guys"
what pissed me off the most about that thread, though, was the idea that statistics is a fully automatable field
uh
This is as they're laying off a bunch of people, and saying it's our mission to become an AI company
the point of the job is not to calculate the p-value, it's to 1) design the experiment so you end up with data that is usable, 2) justify all the assumptions necessary to run the test that calculates the p-value, 3) explain what the fuck is going on, in context, accounting for nuance. a computer can't do that for you
it sent me there and it has a good SAO one :O
huhuhuhuhuhuhuhuhuhuhuhuhuhu see what I mean?
im using this search thing forever now
https://developer.mozilla.org/en-US/docs/Learn/CSS/Building_blocks
this covers most of the css stuff. DOM hierarchy is an HTML thing though.
deviantart is a strange place, but they have so many rainmeter skins
bookmarked
(I fail to see a reason for this useful thing to not be pinned)
(who knows, someone might wander in here one day wanting to make a steam skin or smth)
oh right steam uses electron now
seriously though MDN is THE reference for front-end web technologies of any kind
given ive been trying to TURN A STUPID THEME INTO DARK AND ITS NOT WORKING RAAAAAAAAH
oh i know someone who's been working on an extension for a website and wanted to make a recolor but the site... doesn't use css variables. so it'd have to replace every occurence of a color manually in the code. please use css variables
memories of sass/less to compile down to old css for old browsers
thankfully IE is more or less dead and everyone else supports css vars
oh gods why
it's cohost to be clear
now that's no excuse
the trial i've been on as a GRA for the last five years is coming to a close, and the extra data collected for the experimental arm was all done and managed by a CRO
oh, CROs. what are they even doing over there
someone on the team is probably very old fashioned
i've been told i get to wrangle what some late-career PIs have described as the messiest data they've ever seen
oh no
probably a lot of extremely time-consuming, manual checking of cells
i'm looking forward to it as a........ meditative process, let's say
since the only other things on the burner are my literal dissertation and screaming about having to find a job
oh yeah... job hunting right now sucks. good luck <3
Oh god
fwiw I've had ~5 upgrade problems in ~3 years of using arch, only one was significant and I got a fix by making an arch forum post after 1 day. The rest already had forum posts with answers
The whole 'constant updates = instability' thing has def not been my experience
to be fair the significant issue was that my entire desktop was very broken in a weird way
in my experience using non-megacorp/open source/free software requires spending time figuring out problems - with big tech things you're paying for them to figure that out for you
so far, doing so has been worth it for me, but it's not for everyone (I say this non-judgmentally, wanting stuff to just work is valid)
I use arch for my laptop, desktop, and home server lol
The only issues ive ever had is intel has twice broken their own drivers, but it was fixed within a day
And i just had to use the LTS kernel and went on like normal
Im not sure id recommend it for beginners tho
yeah I think if anything I have forgotten what it's like to be a beginner
I mean there are absolutely lower-maintenance distros for linux as well
popos is really popular, as well as ubuntu
My biggest issue (though granted this was literally over 10 years ago) was the need to not update too infrequently, lest you asplode a thing.
And now I want to switch back to Arch...
i had this experience with rolling-release distros, too, and at about the same time, so i haven't used one since
Yeah, it's definitely not specific to Arch!
pop os may be good, but i will never use it because it is called "Pop!_OS" and i simply cannot. i cannot trust an organization with taste that bad
lmao
(i fully understand how petty this is, do not @ me)
valid concern
I still just feel like Debian is a solid default for most people.
I do miss Slackware, though...
Yeah you can run into issues if your keys are stale
Hey there!
Does anyone know how to update a mod? ^^°
I really love the Social Network mod, but it's not working with the 1.6 anymore. Could anyone maybe explain me how to do it?
The creator isn't active anymore and I really miss that mod in my game so bad.
I'm not skilled in programming at all, but I thought maybe it's not too hard, as I would just try to update it >.<
It's this one:
https://www.nexusmods.com/stardewvalley/mods/8255?tab=description
It me
I want my things to just work
#making-mods-general this channel is for off topic
Nope, but if it's not too hard I thought it could be possible to do it. Sorry if the idea did sound stupid
we can talk about it, just in the other channel
okay, sorry! >.<
yes
this is the one I get after not updating for ~months, so I have a fixkeys shell function for it
Looking the code, doing some stuff, thought, This is a very specific number.
if (monster == null && this.generationRandom.NextDouble() <= 0.20000000298023224)
{
monster = new HotHead(Utility.PointToVector2(spawn_point) * 64f);
}
like, why not 0.2???
it's probably the nearest approximation to .2
because it's not possible to represent
yep
banking software worth its salt uses fixed point rather than floating point
Or Decimal
isn't Decimal just a particular flavor of fixed point number?
It Depends TM
oh no
yeah, apps working with money don't use double/float at all for calculations
and if they do, i'm out
can confirm, I don't think we have a single float in our entire codebase at work
it's all decimals and integers
decimals in the database too
and we still deal with the nightmare that is rounding errors every day
but at least with don't have to deal with floating point errors
If you're working with currency and you don't have a good decimal type, just use integers.
yep. store cents
If you ever use float for currency you are officially crazy
this somehow reminded me of the time i was trying to do some C(++?) for Nintendo DS homebrew, where floats were super slow, so it was recommended to use integers instead. every 8 a single pixel on screen, for example
storing cents does work as long as you're not dealing with something that should be able to handle partial cents like interest rates
Thanks I hate it
you could in theory scale up the integer by another factor and still use an integer
which is probably the performant way to do it
using a decimal type is easier though and has more tools available out of the box
especially when dealing with rounding error differences that you have to redivide across the transaction for something
Using a decimal type is just using integers with the compiler doing magic behind the scenes.
I can't understate how incredibly annoying rounding errors are to deal with in a financial application
All the fun of a rounding error but now you have the kind of shitty people who only care about money whining about it
All currency should be measured as a bigint with satoshi as the unit /s
especially annoying are things that have to be divided among several items in proportion to their value. Like if you get 150 loyalty points for a transaction but then you return one of the items in that transaction, some of our clients want a part of your loyalty points to be taken away again and then you have to know what proportion of the transaction was returned. And even more fun that if they then later return the rest of that transaction that you don't end up accidentally subtracting one more or one less loyalty point that they originally got due to rounding
all currency should be unsigned longs and measured as a decimal percentage of total money that exists
there are no possible problems this could cause
I'm glad I don't deal with financial software. I have a couple friends who work at Squarespace and some of the stories... yeah no thanks
SquareSpace, you mean the new... owner (runner? Host? Manager?) of Google's custom domains out of nowhere?
uuugh dont remind me
i've worked in such software a bit... we had a backend, an iOS app, and an Android app... and we've had taxes, tips, split payments, etc to work with
and of course the calculations were getting out of sync between the platforms
the amount of bugs we have where its just that we have breakdown of the amount of tax someone paid in each different tax bracket but then due to some of the calculations not being available cached we have to recalculate and accidentally end up with 1 cent not being accounted for. Then we have some customer very angry again that their invoice's tax breakdown is 1 cent off from the total (even if the actual totals given are correct)
At least it was a good reason to migrate my domains to Cloudflare
Why did the internet ever leave Geocities? That was enough for everything imaginable
Cloudflare is also an evil megacorp 
I use cloudflare soz
welcome to the horrors of Web 2.0 3.0
All currency should be expressed as the number of atoms of gold /s
the corporate internet must die
I use github and now squarespace
You're not gonna have a registrar that isn't a corporate thing
the solution that was used? fucking shared JavaScript code on all 3 platforms
evil megacorps are hard to avoid, because they worm their way into everything
At least Cloudflare already handles my DNS
javascript??? for payments???
Because JS famously evaluates the same everywhere
Return to neocities
BEGONE FOUL DEMON! THE POWER OF YOBA COMPELS YOU
Domains are the one thing I wouldn't trust Cloudflare with in a million years at this point. Everything else on the free plan, sure.
reject SEO, return to neocities
that was 10 years ago, i hope they changed that logic since
seo has ruined every single internet search already
My domains are mostly on iwantmyname right now I think? Just one on Cloudflare, formerly on Google
I just like Gandi personally.
what'd they do?
we are moving more and more to a central API that handles all that kind of business logic which has absolutely helped but our report generation for stuff like invoices is old and still fully reliant on stored procedures and they constantly have little rounding errors and issues
we needed our stuff to work offline too
our cash registers are the same unfortunately
they froze all the domains of the $120k bill company didn;t they?
If nobody else gets to it before I get to my computer, I'll grab you the links!
cash registers need to continue working when the internet dies and if the electricity dies it needs to flawlessly come back online later as if absolutely nothing happened
it's a difficult problem
it really is
smart
I dont think i have anything tied to cloudflare anymore
Wait, I'm supposed to leave Cloudflare? I have them hosting all of my domains currently
i have no clue how to move a domain somewhere else anyway
I have 8 domains, which I actually moved from NameCheap to Cloudflare
i have Google Suite set up on my domain, and i'd be so scared of screwing that up
you don't have to, no one can avoid bad companies entirely, me complaining about it isn't to shame people who still use them, more at the state of the world
I'm just not in the loop about why, need to do some searching
Cloudflare does provide some pretty neat services.
do what you can, don't be mean to people who don't
fun fact I can say cause it's not true anymore. For like 12 years our cash registers had a file with all the login details in plain text for every cashier so they could still log in when the internet died
anyway I use proton mail now 😎 it's great 😎
It's just frustrating because big companies are bad.
Same here, I moved all of my custom domains from being hosted on FastMail to Proton
dang i love radicalizing my local farming simulator video game community chat room
I too use protonmail
I have the Proton family plan, and use their VPN, Email, Calendar, and File storage
and once proton ends up being evil I will move once again
such is the cycle of life
How about instead of moving, you just grow an evil mustache and join them?
I was already this way :P
unfortunately I grow a very poor mustache
dang i love being part of an already radicalized local farming simulator video game community chat room
part of me is very tempted to just use a private mail server but I know that a lot of places don't like you not using a major provider
it has a lot of issues yeah
yeah, I was already concerned I'd have issues using proton mail
I have a raspi I could use that in theory
it is really that google doesn't like it, google serves like 70% of all email traffic
your emails will often just end up in people's spam folder
surely there's nothing wrong with wren@<some_ip>? surely that is not at all sketchy? /jk
https://www.reddit.com/r/sales/comments/134u0mq/cloudflare_ceo_publicly_calls_out_sales_team_in/
Short version, the sales teams apparently like to reach out under the guise of trust and safety issues to forcefully upsell people to the business plan.
Email is the one thing that I never went the self-hosted route with. Too much trouble than it's worth.
(Sorry, enterprise plan.)
And I self-host a lot
I don't even use that many things
wow neat
Honestly I'm surprised one of our projects is coasting by on Cloudflare's cheaper plan.
what dashboard/service is that?
I've been thinking of setting up self-hosting for my ebook library and maybe a personal internet radio
It's called Homepage
got a fancy dashboard while my server is just a funky thing I ssh into
AAAY I see calibre in there, hell yeah
It's really simple, you just setup some yaml config to point to each of your services
I did set up a personal domain for email, so if i do switch again i can keep the same email address
Before Protonmail and Fastmail, I was using Microsoft's Outlook hosted emails and then they discontinued that
I have a few custom email addresses, and one specifically for E4ward use for disposable throwaway virtual mailboxes.
I'm involved with a project that sees about 100 TB of bandwidth a month, but it's across 10 billion+ requests, and Cloudflare seems just fine with having us on the basic pro plan. I know it's not a ton in the grand scheme of things, I've just expected them to be annoying about it because... publicly traded company.
Every company -
Phase 1: Do no evil
Phase 2: Profit
Phase 3: Evil
massive distributed cat picture sharing service hopefully
I'd... recommended having some fast migration plans in place just on the off chance this becomes more common, though.
I have prepared a simple flow chart.
Yes.
Basically
Its weird how i cant delete custom emails with protonmail tho, ive got a few i dont really plan on using
In unrelated news, raspberry pi is publicly traded now

oh no
profits were a mistake
WHAT?
I control the subdomains, i could do this...
Take that back! 
do it!!
I think it happened this week
Good news: Raspberry Pi doesn't have a monopoly on lil computers
Yeah, was about to say that. They're no longer the only game in town.
Honestly you haven't been able to buy them at a reasonable price in forever anyway
Feels kind of like little of value was lost honestly
There are arguably better options than Raspberry Pi that just aren't the brand name
They have some of the best arm support, but yeah theyre not really cost effective
dang this is awful. did they ever respond?
At least Nvidia wasn't able to acquire ARM, that would have been very upsetting
Honestly all my little hacking projects lately have been on ESP32-S3 boards too, instead of using RasPi's
the CEO said.. something on twitter iirc
I am so ready to be able to buy a Snapdragon based system to run all my home server stuff on.
Speaking of which
If someone knows of a board/microprocessor that has zigbee support, that would be great. I use zigbee for a bunch of smart home stuff, and i want to be able to create my own things without having to put them on wifi
But i havent found anything
Does an ESP32-C6 not support zigbee?
Yeah, a C6 should.
The ESP32-C6 SoC (System on Chip) supports Wi-Fi 6 in 2.4 GHz band, Bluetooth 5, Zigbee 3.0 and Thread 1.3. It consists of a high-performance (HP) 32-bit RISC-V processor, an low-power (LP) 32-bit RISC-V processor, wireless baseband and MAC (Wi-Fi, Bluetooth LE, and 802.15.4), RF module, and numerous peripherals. Wi-Fi, Bluetooth and 802.15.4 coexist with each other and share the same antenna.
Oh maybe thats exactly what i want
You can get a dev kit for like 10 USD, they are very low power, programmable via USB.
they don't run linux right? only rtos's?
Yeah.
Yeah.
I did it
If you're used to arduino, then you're set.
same as an arduino, yeah
I can't wait to get into more home automation stuff. I've been researching smart light switches, but I really want to wait until there is a good Matter/Thread one.
is zigbee worth it over wifi? all of mine is wifi and works fine
I'd imagine so.
If I didn't care about having my light switches augmenting my thread network, I'd probably just get Lutron switches
Lower power + not poluting wifi
I assume they use less power, but I don't know any other benefits
Granted it's less of an issue with modern wifi tech, but you still want to minimize traffic on those bands for the best performance.
Benefits are not having 100 things on wifi
Not having to use their app once to get them onto wifi
Not knowing who theyre talking to
Your light switches don't need IP addresses and HTTP servers
you can use arduino stuff yeah
even the arduino IDE
I also have my zigbee stuff forward to MQTT, which is can parse much easier
Zigbee is all local
ooh, say more! why lutron? I have wanted light switches for a while but never gotten into them/still use bulbs
@silk narwhal
In a powershell window
# Go to user home
cd ~
# Create ssh directory. It's ok if this exists, don't forget the '.' at the beginning
mkdir .ssh
cd .ssh
# Create your key, just press <enter> through all the prompts
ssh-keygen -t ed25519 -C "GitHub Key"
# Print the public key, make sure it's the public key!
cat id_ed25519.pub
Copy the output and add an ssh key to your github
https://github.com/settings/ssh/new
Ive replaced a few switches
Theyre nicer than bulbs since they have a physical interface still
I've looked into dozens of options, but what I like about Lutron's is they have a tactile dimmer
If/when I do smart home stuff, I will absolutely be replacing switches instead of bulbs.
Replacing lightbulbs seems so stupid given the failure rate of bulbs.
Not to mention that you can then turn off the switch and oops, no more worky
Its a lot easier and cheaper tbf, but theres benefits yeah
My only issue with smart switches is that I'd need a lot of three way switches and, last time I looked at least, they... didn't do that part so well.
Here's a video I watched about it. Tempted to go with this if a better option doesn't come up anytime soon https://www.youtube.com/watch?v=bHxIi79tRG0
Lutron just announced new smart light switches that solve a LOT of the problems with the older Caseta switches!
Lutron Diva Smart Dimmer switches are available at Home Depot: https://homedepot.sjv.io/a1gx1Y
(affiliate link)
DISCOVER MORE
I just spent thousands on Lutron: https://youtu.be/GBWtmJPCyGc
Smart Light Switches- I'm Rethinking EVERYTHI...
Not that I know of! I'll be sure to toss anything at you if I find out they have, though.
I just don't like the unresponsiveness of holding down up/down on the switch to adjust dimming
but i'm not sure i can recommend them
i have 2 of these... 1 is perfectly fine.
the other one... discharges any batteries i give it within a week
the other one has been working for half a year if not longer now, and is still on the same batteries
That looks cool, but batteries are a deal breaker for what I'm looking for
i really need to message their support, but i just never felt like doing it
I'm trying to replace the existing switches in my house
I wrote a blog post about my smart home, I completely forgot
My only problem with Lutron is it needs a Hub, and I'm trying to reduce my hubs
yeah... i have two hubs right now
Home assistant
Although I suppose, once I set it up, I'll no longer be using the Hue hub
Remove hub
Lutron hub integrates with Home Assistant, but it's still required for their switches
Philips Hue hub, Flic hub... and then my server that runs HomeBridge
I still want network-controlled lights that I don't need to worry about being on wifi.
They don't use standards, it's proprietary
Then what does the integration do?
Should the key type on github be authentication or signing?
Gotcha
I do like the idea of a rotary switch a lot. It works great for non-smart dimmers. But having one that needs batteries is ridiculous.
The Lutron Hub makes their switches controllable from Home Assistant, Alexa, Google, HomeKit, etc. but it's responsible for communication between the switches.
@silk narwhal The stuff from git config -l is all the configuration options for how you use git on your computer, like who to mark as the author when making a commit (name + email)
The credentials come from the actual protocol you use to access the repository
If you cloned with https://, then you'll be asked for your username and password
If you cloned with ssh:// it'll try to use your ssh key
(btw you'll need to update your remote for the repos you use if you start using an ssh key)
In your directory where you have the git repo:
git remote --set-url origin <the ssh url not the https url>
Ahhh, okay
fwiw, it's the only solution that works for me, since i don't own the place i live at
Yeah, that's understandable. I am fortunate enough to own my home while a lot of people don't.
Fuck landlords
My favourite thing about my git config: commit.gpgsign=true.

I own a home now, but I've done "apartment hacks" prior to owing a home
@itsfolf Break the law and put our IP space at risk and we will protect our customers by forcing you onto you using private IP space, which costs a lot because it’s bespoke.
Wow look at this overachiever who took the time to set up signing for commits
I'm in the slightly less lucky but still absurdly lucky camp of not owning right now, but still being able to do... a shocking amount of things.
Yes... i, a renter, have not replaced my light switches
I don't remember what computer it's on
No sir
Like, I've gotten away with anything reversible that doesn't cause damage at different places
@gaunt wadi, holy hell. That's the CEO responding right there.
there's a reply after it but you need to be signed in to see the thrread
does the money really alter your world view that hard
I think it's more that it's mostly the type of person who doesn't need to change to become that who becomes a CEO of a massive corporation.
the one remaining nitter instance that I know works: https://nitter.poast.org/eastdakota/status/1797489379411394969
Rather than becoming one changing people.

Oh you beat me to it
I absolutely understand needing to force people onto a private IP in that situation to protect all the shared IP users. But I don't understand why it's so damn expensive for Cloudflare to give you a private IP. IP addresses are effectively a software thing and they already have crazy routing tables
Okay, minor snag, can't seem to figure out what the git repo directory should be
Yeah, requiring BYOIP there is 100% fine.
But the way they went about it is... horrendous.
Having a private IP should be like a... $5 a month thing not a $120,000 thing
The requests are going to the same machines anyway just with a different number
oo ty
mhm
Wherever your mod is, that’s the git repo usually
Mm, I'm getting either fatal: not a git repository (or any of the parent directories): .git or error: unknown option 'set-url' when I try stuff that I think are the git repo directory
Sounds like its not a git repo
Oh wait I typed it wrong
The git repo is the directory with a .git folder
Ahh gotcha
So probably wherever it says that you typed it wrong is right and I’m dumb
neat ty! how do they work if you turn the light off programmatically? does whatever state the switch is currently in become off, and toggling it becomes on?
Going off the adage that no news is good news, seems like it worked! yay
i.e. in regular switches down is off and up is on, but with those it can change?
git remote -v
The up/down do not physically depress, they're more mushy. So they press but then come back up.
ahh I see
so more like buttons
The rail part is like a hardcoded presetting that you can override with the app
Seems right, it's got the url I put in
Yahoo
It has a traditional look that fits in well without looking too smart, which is something I like about it
Then just git fetch —all to make sure things work
This reminds me that I need to make a trip to the electronics end of the old fort commercial area here and buy one of those, lol.
Is that two dashes that got auto-corrected into an em dash
I'm getting something about the authenticity of the host not being established
Ah, looked it up and it seems fine
If you don't get that for the very first time you ssh into anything, that's when you should be slightly spooked.
... This is a lawsuit waiting to happen.
I still just have the first Pi!
And yes, it absolutely bloody is.
The only pies in this house are the ones you can eat, you see
I've been meaning to mess around with a raspi for years though
I don't even know where to begin with how fucked up the whole thing is!
As long as you typed yes and it works you’re good 😊
Yippee!
Okay, next I'm assuming is doing this stuff #making-mods-general message, I don't really know what my_pr_branch would be though
Whatever branch you’re on in git status
Don’t run push -f until you’re sure everything is ok, I don’t want to be responsible for losing all your work lol
The f is for force
Everything before that is able to be undone
It's fine! I've got the actual fixes on a local copy and don't care about the readme
Ok cool lol
I made the changes to the readme on the github website itself, so it looks like my local stuff is still on the commit before changing the readme
Big success!! Thank you so much, now to hope I don't forget all this next time I screw up ahaha (the messages will still be here I suppose)
Dw about it happy to help 😊
Git is just one of those things that makes sense to me
Also for the future you don’t have to make that temp branch, just wanted you to have a backup in case something went wrong
seo rant: i look up words to get the definition and i get fuckin brand names for things
makes me angery
i would not like to include "define" in my search query or visit a site for definitions
I want products and brands to be abolished
thank you
not an seo rant but sometimes when I search something for a game I'm playing I forget the word is totally normal and then none of my results are related to my game.
Like I'll search "clockwork" with no further words fully expecting to be given the runescape wiki page for clockwork
Here at AquoSoft, we understand your frustrations. We too experience human emotions towards things on a daily basis. That's why we created AquoAI™️, the all-in-one solution to literally everything. With AquoAI™️, all of your problems will be solved by leveraging the power of the internet in its infinite wisdom.
incredible
what does a world without products even mean
just... nothing
nothing that is produced in any sense of the word
I'm glad you asked Crumble. With AquoAI™️ tangable objects are a thing of the past, because you won't own any. All of your physical items and cherished memories will reside inside the Cloudova™️ where it can be harnessed for all humanity to enjoy
Sure thing citizen! With the power of AquoAI™️ your wish is our command!
Processing...
Processing...
Don't be afraid! With our trademark Blockchain Metaverse technologies you'll never lose the things you own
That's amazing
youre an animal
Woah there pardner.. not my doing. Colleague's shenanigan's.
I'm not a monster
My favorites are the ones with the 'a' for - ostensibly - alpha that also are a beta or final
Plus using dates that don't keep them in any kind of natural order
I’ve seen worse tbh
My grad school collaborator liked to forget what his folder names did, and he never put dates
ISO dates or don't bother imo
what do those numbers even represent
I thought dates but that makes no sense
oh wait is it american date format
why would you ever use the american date format for files names man. Use the ISO format cause it would correct be ordered by date even when sorting by alphabet
in this case 080125 for the top one
2008 January 25
The best versioning system needs a few "Filename (Copy) (Copy)" in them as well
_1
I'd assume they're dates from 2008/2009...
idk what else they could possibly be
and they started at 1 am
I mean MMDDYY does sort by date within a year
I’ve considered changing my folder date system to sort better, I’m using MM/DD/YY like an American heathen
Year Month Day or death
YYMMDD or YYYYMMDD the only two options
Crumble
I'm gonna make my system do YYMMDD or MMDDYY or DDMMYY
But randomly
Chooses a different one each time.
md5(MMDDYY)
at that point you might as well use uuid v4 for your names 
not uuid v6 cause that actually encodes a timestamp
nor 7 I suppose
Just use epoch time in base64
Yes, from 2008. I have to catalog someone's files so we can get it under actual version control. And this is some of the best I've found so far, believe it or not, where they're all in the same folder with somewhat of a versioning format.
Plenty of other codebases where there's just endless copies of folders with nothing renamed; just some differing modified dates
Just throw a folder in as a commit, then the next one as the next commit, let versional control figure out its own stuff
Idk if you can spoof the date tho, that would be neat
I had to do a lot of funky git stuff today
I both did a git revert and a cherrypick
unrelated to each other
weirdest git thing i've ever done is delete a file from history
strong desire to have smart window shades now
ignoring the fact that I don't own a home/cannot install them 
they're not thaaaaaaaaat expensive in the grand scheme of things
but yes they are expensive
me justifying smart shades to myself
So long as you put the old ones back...
They go up
close automatically ig
wow saving yourself a 3 second excursion to a button
my mom does have some shades thing that detects when it's really sunny and then automatically goes down
that's nice
but I assume that's not what it means cause smart seems to have become synonymous with connected to your phone
I would set them up to open/close at certain time of day, have phone control, ideally detect brightness
Right now, I have a button that if you push it, all the lights in the apartment turn off
its great
this exactly I do not want to stand up and walk 2 feet to the shades I want to wake up and have the shades rise like a god amongst men
Stand in living room, press button on phone, and all the shades open in unison
light shines down upon you like the God you are
presence detector: I walk into my office and sit in my desk and all the shades open
the universe beckons to my call
my whims are reality
I got a presence detector, the lights in my office come on now
It was like $7 at ikea
oh shit
this is one of those things where I am of the boomer mentality and just want it to be its own separate thing with some physical switch/button I can press. I don't want smart lights or smart blinds or a smart home heating system or all that stuff
don't boomer yourself people have different interests without being boomer
the secret is i dont use any of its advertised features
also the beauty is that you have physical switches too
I'm the one giving it instructions on when to do stuff
when do the lights turn off again
when i press the lightswitch lol
every time aquo blinks, to save power, and then on again when his eyes open
ah yes
i didnt get around to programming that
power bill: $0
tbh I don't know how to do it
Yeah, it detects the room is empty if i dont move enough
so instead its just pointed at the door
oh my god they're so cheap https://www.ikea.com/us/en/cat/electric-blinds-44531/
we used to have presence detectors for the lights at the office and without fail it'd turn off a few times when I was there cause I don't really move while at my desk
so I just gotta start waving my arms or something for the lights to go back on
If i actually had a house i wouldve done it
but i dont wanna mess around with the apartment shades...
buying a house to have somewhere to install smart shades
plus if i move who knows if the windows would be the same sizes
yeah, same, they're pretty in the wall
for blinds just give ma funky chord I can pull to move it up or down it doesn't need access to electricity at all
pure mechnical
ive got...
is it this @pliant snow https://www.ikea.com/us/en/p/tradfri-wireless-motion-sensor-smart-white-60377655/
it does say last chance to buy/assume it's no longer in production
$9
I use home assistant, with a zigbee dongle yeah
hot damn
you'll need a USB antenna thing to have zigbee support, but they're like $30 and then they work with everything
I've got...
9 zigbee switches
8 plugs
motion sensor
door sensor
button thing
air quality sensor
Mine is thru aqara, i dont necessarily recommend it, it really likes to disconnect and not play well
I think ikeas got one, but i havent used it
I used to use the Nest Secure door sensors 
oooh they do
Those were discontinued, but otherwise worked well for me for years
So now I'm in the market for a replacement
Also had them on my windows
The biggest brain move I made was some sensors are only one way and usually require some proprietary hub to work, like weather sensors
I have Ring on my windows, but I want to replace them. Looking for something I can self host without a subscription.
turns out all of those typically communicate on the same frequency band, and if you get a USB antenna, you can just pick up their signals
its great
at my old apartment, I used to pick up tire pressure sensors as they drove by, since they also used the same frequency
with six dogs ™️ you can be notified of any intruders with no monthly subscription *food not included
The holy grail of home automation is having accurate presence detection that knows who is where
it's also mildly creepy
But like everything I can possibly want to automate is only limited by my ability to know where I'm at in my house, and for it to be responsive enough
the holy grail of home automation is being able to turn on the ceiling fan without getting up
if you want to impress the lady in your life, give her a button to turn on a ceiling fan
what zigbee dongle do you have mr quova
itssss
I'd feel rich just having a ceiling fan
Users can add and manage ZigBee sub-devices through this device and smart home platform by plugging ZigBee 3.0 USB Dongle Plus into PC or Raspberry Pi and configuring accordingly to achieve communication between devices connected to the platform and create more intelligent linkages for home autom...
If only I was a complete nerd and wrote an actual blog post about my home set up
It needs more links to the actual hardware tho, I might add that later
We need a smart home enthusiasts channel
I do many things I dont advertise well lol
Yeah I'm gonna add more pictures and links this afternoon tho, its a bit sparse rn
neat I will be eager to read it
i have planned to blog more for a while but life gets in the way
I really like the YT channel Smart Home Solver. I get a lot of ideas from him.
maybe also get rid of the farming game stuff channels, it's kinda off topic
I did buy the RSP32-C6 as well, hopefully I'll be able to write my own Zigbee firmware for stuff
@ green beans
I wondered why I didnt know about it before -- it only got support in the arduino libraries 2 weeks ago lol
you'll be able to avoid my grumbling about them 🙏
Can i make a thread
no
dang
can i give myself the permissions to make threads...
I think so
ugh no
can your bot make one for you
...interesting
tbh just ask the green beans
this sounds like a gross violation of authority
yes
privelage escalation would not go over well, yeah
is that part of the charm? maybe
oh for sure
it is
but i also cant be bothered
a bunch of ex junimo talking about how it'd be fun to violate the rules
sounds fun to make the bots do things they shouldn't, again
whats the worst that could happen
i can't stop thinking about that guy who discovered an exploit that allowed anyone to edit any device issued by their ISP, to any user, remotely
This is truly one of the craziest scenarios I've ever seen. An API endpoint left wide open lets you hack anyone's router.
Thanks for letting me make this video Sam!
Article: https://samcurry.net/hacking-millions-of-modems
Sam Curry: https://twitter.com/samwcyo
🏫 COURSES 🏫 Learn to code in C at https://lowlevel.academy
👕 MERCH 👕 Like the shirt...
that sounds.. not good
and get anyone's personal data while they're at it, i guess
Hope that goes well! I've been enjoying my little esp32 projects. Haven't done anything for smart homes yet, but I really should.
or the article directly: https://samcurry.net/hacking-millions-of-modems
Two years ago, something very strange happened to me while working from my home network. I was exploiting a blind XXE vulnerability that required an external HTTP server to smuggle out files, so I spun up an AWS box and ran a simple Python webserver to receive the traffic from the vulnerable server.
Things like these make me wonder how incredibly unsafe our work software is if anyone knowledgeable actually cared enough to try and find a vulnerability
we get pen testers but I know for certain they keep missing things that I know about
oo interesting
oh that's quite a major vulnerability
well ok I don't think we have this...
that vulnerability is just that they didn't have any authorization or authentication on the api?
@cinder karma I'm about to play with electricity, please tell me how I don't kill myself
incredible
well, major in impact, rather than 'size'
yeah
it's interesting
idk how you end up accidentally disabling authorizaton on your endpoints
we have authorization on even for debug builds
might have just assumed nobody would find the endpoints in the first place
this does make me feel validated in my hesitance to go with the client side framework logic we were doing at first at work
I really want to up my DIY game, but I'm terribly afraid of things that can electrocute me or chop my appendages off
Maybe start with just a lightbulb
What
Please tell me you're not planning on playing with wall voltage
What are you doing
Replacing a light switch
Yeah I think I'm prepared, I know the breaker to switch, I have a multimeter, electrician's glove
And that was the last time we ever heard from Matt 🙁
eLeCtrocutedMatt
speaking of qol features in IDEs
say a prayer tonight to all the statisticians who are within a metaphorical arm's reach of the fda because sas looks like this
it's such an eyesore and you can't change it
I'm glad Databricks added dark mode, made staring at a screen for hours a day at work much more tolerable
rstudio is awesome by comparison but now i'm gonna be checking the forums every few months to see if they ever add the rainbow bracket thing because people have asked for it
public class Program
{
private const int NumIterations = 1000000000;
private Stopwatch _stopwatch = new();
public static void StaticAddTwoRef(ref int input)
{
input += 2;
}
public static int StaticAddTwoRet(int input)
{
return input + 2;
}
void AddTwoRef(ref int input)
{
input += 2;
}
int AddTwoRet(int input)
{
return input + 2;
}
public static void Main(string[] args)
{
new Program().RunTests();
}
void RunTests()
{
RunTestStaticRef();
RunTestStaticRet();
RunTestInstanceRef();
RunTestInstanceRet();
}
void RunTestStaticRef()
{
var num = 0;
_stopwatch.Restart();
for (var i = 0; i < NumIterations; i++)
{
StaticAddTwoRef(ref num);
}
_stopwatch.Stop();
Console.WriteLine($"Static Ref Finished. Time: {_stopwatch.ElapsedMilliseconds}ms. Num: {num}");
}
void RunTestStaticRet()
{
var num = 0;
_stopwatch.Restart();
for (var i = 0; i < NumIterations; i++)
{
num = StaticAddTwoRet(num);
}
_stopwatch.Stop();
Console.WriteLine($"Static Ret Finished. Time: {_stopwatch.ElapsedMilliseconds}ms. Num: {num}");
}
void RunTestInstanceRef()
{
var num = 0;
_stopwatch.Restart();
for (var i = 0; i < NumIterations; i++)
{
AddTwoRef(ref num);
}
_stopwatch.Stop();
Console.WriteLine($"Instance Ref Finished. Time: {_stopwatch.ElapsedMilliseconds}ms. Num: {num}");
}
void RunTestInstanceRet()
{
var num = 0;
_stopwatch.Restart();
for (var i = 0; i < NumIterations; i++)
{
num = AddTwoRet(num);
}
_stopwatch.Stop();
Console.WriteLine($"Instance Ret Finished. Time: {_stopwatch.ElapsedMilliseconds}ms. Num: {num}");
}
}
these two screenshots were taken on two different monitors, sas would look extremely fuzzy on my desktop


