#💀 Cortex Engine
1 messages · Page 9 of 1
wtf are filters
might be a new haxe thing
yeah... in a new project with just haxeui and ceramic in it. I get about 3s recompiles its pretty okay
i just saw daz's sub 1s refresh and got a bit jealous x)
seems to be related to DCE? could be wrong
Run ceramic haxe server and it’ll use a server to build and make it even faster (but there can be bugs there too thank you haxe)
https://github.com/HaxeFoundation/haxe/pull/12087
https://github.com/HaxeFoundation/haxe/pull/12085
https://github.com/HaxeFoundation/haxe/issues/12109
that is what I think atleast looking thru PR's
here is a recent more expensive project. I blow most of the time in macros
these numbers dont add up
yeah wait wtf
lmao
in the lil test project I have its just the haxeui buttons demo and a quad in the main scene
nothing big or heavy
i wonder if it could be resources... not resources
Gonna go with HL for now
I remember hearing a few things about a haxe compilation server
Where do i find more info?
its on the haxe site iirc
¯_(ツ)_/¯
both have upsides and downsides
personally I think HL's downsides farrrrrr outweight hxcpp's downsides
especially with cppia, hxcpp is kindof a no-brainer imo
the beauty of making your own shit is that you can always add or change it later tho
Only downsides i see with HL are lack of libraries and no cppia
hxcpp's compile speed is abysmal
which should be fixed with cppia
and HL has questionable support at best
externs for hxcpp are also way easier imo
I already learned how to make HL externs im not gonna give it up now
Questionable support or not, compile speed is really quick
I might add hxcpp support later for other platforms (e.g. web and mobile if HL doesn't perform well in those environments)
Peformance might not be the problem
But building for Android with hxcpp is painful enough as is
HL? Won't be better
And hxpp gives you more freedom overall
With HL you have this proprietary VM that only a handful of people actively maintain and also very few people are truly knowledgeable about it. Meanwhile hxpp gives you the freedom to run on many platforms and kinda do whatever
As far as I’m concerned, I don’t NEED cppia for Ceramic because I have the web target, but it would be nice to have it at some point
HL support on non-windows os is also not very good
No reason to use it when there is hxcpp
Not saying few people working on it is bad, but it means that things that are not relevant to those people are likely not gonna work and resources about how to use HL effectively is scarce
This
If you support both cpp and js or cpp and cppia then I personally see no reason to use HL
Thanks to ceramic’s electron integration, js and cpp output of Ceramic behave almost like native, with a window that looks identical and same file access
Indeed
I'd still like to have cppia some day
fun fact to get cppia to work for me i had to remove the ceramic_cppia_host flag
This is just because that flag is supposed to be used with other changes, not surprising
Ah okay
Practically HL for me using heaps is the same as JS/electron in ceramic. Fast compile, don't really care what target it is, don't need to think about asset servers
Similar enough to html5 export on OpenFL and Flixel
sure, makes sense. For me it is also about development and shipping. I wouldnt ship anything cppia and I know at least 5 other persons that have successfully shipped a commercial product using hxcpp. With HL Im not even sure anyone besides Shiro released something commercial. Everybody else seem to only tinker. I'd love to be corrected on that
right! so we know that 2 studios used it. and they used it bc they know Nicolas personally
since Im not a studio and I dont know him personally HL is kinda pointless for me
there are more reasons, but you catch my drift
certainly not arguing for hl over cpp, i'm more of a "whatever works for you" kinda guy
true. I dont wanna come across as a HL Hater, but Im asking how I can take anything HL seriously when compared to hxcpp/cppia
I do think its a serious platform, the number of users who use it are small to our knowlege
but its reach is quite far, that in itself does give credibility
im certainly also in the hxcpp > hl camp
perf is better in my tests, extern dev is night and day
i actually wish the foundation had spent more time / resources on hxcpp than hl, as i think someone else had pointed out (Zeta perhaps) they are both doing the "same thing" in really different ways, so seems like uneeded competition
that is not a "waive it away" kind of figure, even though hxcpp probably does roundtrip that
im not sure commercial success is a good metric for tech stack tbh
i'm using it as a battletest metric
you can write a really fun, good, successful game (or whatever) on a dogshit stack
I have access to a faster computer now, and therefore hxcpp looks better now then HL.
I think they both serve Haxe well, having multiple implementations makes the ecosystem stronger
absolutely, but, if we're looking at stability i don't think i've ever seen any known hashlink game have a bout of angry users complaining about it
yeah, it would be nice if the foundation took additional responsibility over hxcpp with the dropping of some targets
community contributed or not, hxcpp adds a lot to haxe
i kinda used to think the same (that HL had a place, and 2 VMs are better than one), but now i actually sort of feel its fractured effort and given limited resources, i would certainly prefer a bulletproof hxcpp than hl - but as someone mentioned, above, its not going to happen because shiro is the driving force
if the externs system was better, i would be more sympathetic to hl probably
You know what i've realised a lot in the foundation in idle discussions
they really seem to have little idea how "deep" haxe stuff really is
yeah, and what's available
right
it's a bit ironic
Well the foundation is a group of people, who looked the most clueless?
... ... ... they should read the roundup - although - Rudys here and hes always pretty engaged, etc - so they must have some idea
rudy is engaged but i'm talking people like simn and nic
i don't really know if rudy casually chats to these guys about haxe stuff
their primary exposure to "haxe stuff" is pretty much shiro and whatever sponsors are around (if there are any)
Not bad i guess, just probably contributes to the reinventing the wheel
altho i'm sure nic would just make his own version anyway
seems like the haxe way x)
Honestly I'm kind of a fan, having compiler and tooling close to the ecosystem makes it better overall
I think what the haxe team is doing is cool, however I don't think it is the best solution that could have been done
HL is cool but I genuinely don't see the point
Spending more effort on JVM is a better ROI (time investment)
That's my opinion atleast
haxe seems to be setup in a way where you just work on what you're interested in
then when times start to prompt a big focus, they put that effort into making haxe better rather than targets
Well the targets get updated if underlying functionality is required to take advantage of the Haxe changes
hashlink bytecode this, cppppia that, everyone just needs to do what kha did that one time and embed javascript into their c++ game and compile haxe to js for scripting
Why do any languages exist besides javascript
I compile haxe to js for scripting too lol, but the whole engine is script in that situation
No need to embed a js runtime to native
Just js everything
Anyway, on the topic, it totally makes sense to me that when you started creating a tool like Haxe, you value more in-house custom tech than « industry-standard » tech. So from that perspective, making HL is following the same logic. You know you won’t ever be blocked because « jvm can’t do something you need » and so on. Instead, you can shape everything in the stack yourself
hxcpp runs circles around that number. there are mobile apps/games with 100+ million downloads
my point being you have a lot of different parties using and contributing back to hxcpp over the years
but again, I dont hate HL or wanna bash it, I just dont think it is there where it wants you to believe it is
and when it comes to usability, cppia is so under-documented and different/low friction/integrated that it baffles most people that look at it with the expectation to find something like a lua scripting setup
TBS, I will write some docs soon
what do you mean by this? That its super easy to use but people arent aware of that?
yeah, and that it is also super easy to use it wrong
right - gotcha
HL's extern system is pain
Pure, documentation-less pain
yeah, i also just dont like the system itself, that you have to build that sort of "glue" code to box things between the VM and the lib
It is annoying but i think that's just because of the design - haxe doesn't map to C well
And
Stare at the hlsdl sources long enough
And you'll get the grasp of it
I think its fine, many ecosystems thrive because of it
I think the main problem is not the glue code itself
Rather that you have to do double the work
Externs on haxe side and the glue on the C side
i mean, thats what i meant (what else could i have meant?): its double the maintenance. I was going to port hxWidgets to HL at some point, played with the externs, it worked(ish) - but the burden of having to maintain two codebases (plus the hxWidgets itself - and for what reason?) was a "naw, im out" - i would imagine any large lib would be similar (there is scope for generation here ofc, like ammer, but im guessing its not as simple as it initially seems, given that ammer afaik is no longer in active dev)
Yeah defo
should see if hxwidgets works on cppia 😆
it should do, with relative ease (compared to HL)
just need to make sure "all the bits" are there and that should be it, the externs should work as normal
I didn't know HL was so hated
how is hxcpp better? (the only targets i used were --interp and --hl)
i loved HL because of how easy was to get an OpenGL context
they're not hating, they're stating preferences
and being objective
it doesn't mean one shouldn't use hl
you don't need to if it works fine for you
if you ever run into a limitation of either target, you'll know more
coool
hl is still a decent target
it's just the new kid on the block
hxcpp is way more battletested
why did HL come to fruition then? i like the coming hot reloading tho
Oh.
These guys are smart so they can get away with doing what they want to do
i like that statement
because of how true it is
the whole crew at shiro is insane
Nic obviously has some vision for VMs
hashlink may not be as literally tested as hxcpp, but nic also has the neko experience
it takes "something" to launch a big company game on a "new" platform and not tank due to stability issues
i guess it was the combo of shiro and motion twin that gave HL the boost it needed.
quite a high risk endeavour
look at no mans sky, i don't even play games much and remember how big of a failure that game was at launch
not their first rodeo since they operated on tech islands for most of their existence. just check the insane amounts of webgames that happened at mt
they even built their own minecraft in space... in flash 😄
https://www.youtube.com/watch?v=XRQ6XE6K-wQ
Old trailer from a canceled game ("Galaxy 55") back in 2012, at Motion Twin.
oh wow 😄
MT have been very prolific for a long time yeah. They definitely know what they are doing
HL is alright imo, it's just focused on their needs, so if yours don't align with it, well, no luck
The fact that I'm mainly working from mac has been making HL toolchain quite cumbersome for me, it always lags behind compared to windows support
And even without that, when you use heaps, it's been many times where you needed to use bleeding edge HL to use Heaps, sometimes it's not easy to keep up when you are outside of their own context
Back before Ceramic existed, I considered using Heaps (and HL) but as my main platforms at the time were mobile, didn't dig further. Building for Android/iOS with Heaps was definitely not an easy path that is well supported
(I wonder if this has improved since then)
Nah you still have to use a random fork of heaps to compile to mobile
zyheaps or smth?
Yeah that one
i went looking theres like
5 android heaps libraries
with heaps having a "WIP" official one (no commits in 7 years)
HL/JIT is only on desktop
for literally every other target it HAS to be HL/C
Web, Andoid, Non-ARM
i wonder what the compile time is gonna be like
Hey daz, I have a question for you. I'm currently looking at shader cross-compilation and I could go for the Spirv-cross/glslang combo but I am down to experiment a little and I have a feeling figuring out how to build the whole chain of glslang, spirv-cross and spirv-tools might become a bit painful.
So here is my idea: repurpose #1338080465642328134 to emit glsl (specifically ESSL/GLSL for GLES2 and GL 3.3+)
I know there are alot of weird little things with shaders and that it is probably wayyyyyyyy more hassle than it's worth. And it would never be as good as existing solutions.
My question to you: Have you tried to make your own shader cross-compiler / language before, if so what was it like?
And the second question: Do you know about other good alternatives to Spirv-cross/glslang? I have looked into a few other options like glsl-optimizer (based on Mesa's GLSL Compiler) and a few other projects here and there. If possible I'd like to avoid calling external CLI tools.
To conclude this message, if I do go through with this I might regret it later, but that is part of the fun 😄
short answer: No, not really. Was never interested in designing a language/compiler. I wrote some stuff for creating AGAL assembly shaders but that was mostly basic pre-processor stuff.
I guess one of the more interesting/esoteric projects is HxSL.
I see, what about the second question? And do you think the idea of writing your own (basic) shader cross-compiler (syntax and feature level of possibly GLES 2) is worth it and/or feasible.
I did take a look at that along with HGSL (from saharan, creator of Oimo physics) https://github.com/saharan/HGSL
But It would not be ideal for me considering the fact that I use IntelliJ Idea (and the plugin for it) which has no macro support.
personally I have no interest in the whole shader language topic. I guess thats bc for me there is little creativity involved and it boils down to an unbelievable grind. But if you like messing with languages and optimizations it's definitely a cool use-case for your own language/compiler
Rob has his own Kong shader compiler thing
Hxsl seems cool, but apparently it’s quite tied to heaps so it’s not easy to set it up for other tech. I have been wondering about what solution to adopt myself for Ceramic, and a shader cross compiler that takes glsl as input seems the most pragmatic approach. Cli tool is fine for me. Same as daz, not particularly interested into reinventing things in that field, just want a reliable and portable solution
I did take a look at that just now, it would probably fit my use-case
I see, that makes sense
Ah nice, but I think I want to stick to glsl, something I know. Just need to transpile it to other stuff (unity shader, mostly, as ANGLE is bringing glsl at runtime already on mac/win)
Taking a look at this: https://github.com/alaingalvan/CrossShader
Looks like it could be an easy-to-hxcpplinc lib that could be used from my ceramic c++ cli
(But maybe it’s better to directly use spirv idk)
Might try something with glslc (google) + spirv-cross at some point
You got so much dedication for this man, those ui transitions are so satisfying
something i dont really get
the format library
who would read PNGs directly in haxe?
why not use a native lib?
I used it in cases where I didnt have a decoder available and needed a quick parser. But it's a tad slow compared to target specific implementations
gah, im really trying to like gimp, but what the fuck man, I place a font element, then I wanna select a font and it's not a dropdown but a textfield and if you type searching your font, it displays a 3km high hovermenu tooltip WITH ALL FONTS it found, no scrollbars ... I dont even see the result that is closest to my search
also it doesnt remember on which screen it was started/closed the last time and always moves to the last screen in your display list. Why is it so hard to spawn on the main display
gah, time to uninstall once again
you don't see the utility in a framework agnostic image library?
i see why it'd be useful but
haxe's kinda slow
compared to C/C++
what does slower have to do with being useful
typically you'd want programs to be fast
slower doesn't mean unusable
a bike is slower than a car, doesn't mean you must use a car
Haxe can perform very close to C++ if you write it with that intention
But of course regarding png decoding, c/c++ libraries might take advantage of further optimizations
You can always use a native c/c++ one and fallback to a haxe-written one if you want to target languages that don't have a native png decoding implementation
As long as it's fast enough for purpose, it's more important to me that the program is maintainable and if needed portable
Fair
Sometimes you don’t want to think and format is pretty straightforward
ok, time to put a lid on this css madness. 3 items left on the list:
- add a custom parser for that
transformnonsense - rewrite the
animationactuator - add sort and mark all 90 directives with their
transitionabletype
dazWebBrowser
dazAwesomeUiApps
Don’t you want to knock at the chrome dev team door 😂
I want to see dazAwesomeGame heh
Fuck G00gle. Had the pleasure of having a casual business dinner with an engineering lead the other day. These guys are soo full of themselves
Just joking of course. I find it funny that Google, as massive as it is, is a competitor on the same thing I’m doing at work with a very small team, and I also don’t fancy working for Google at all
google is evil :3
took 3 questions and he sat there naked. strip away the corpo identity and most of them are pretty.. sad
the usual. google defines itself as slimey mold in terms of company culture, go figure 😄
everywhere, totally chaotic, conflicting agendas but in sum pretty successfull bc too big to fail
but nothing a fine flamethrower cant handle
a bit more tech required before I can show it off
whats the thing that allows you to set a git library in dependencies of haxelib.json
I understand, I really do 😂
your text editor 😅
._.
how do i do it tho
isnt it like
git:gitlink or something
Yeah, that’s not so surprising I guess. I think I would be sad too if I were working there
Yeah, its the same syntax as -lib except without the library name. It's shown on this page under Using haxelib with git versions:
https://lib.haxe.org/documentation/per-project-setup
thank you
it would be useful to add that to the other documentation pages
i think its just a warning in the editor, but the syntax looks right
yeah haxelib didnt complain
okay the warning is annoying me i'll just handle it in the build script
this seems to be the cause of the error: https://github.com/HaxeFoundation/haxelib/blob/development/schema.json#L103-L111
interesting, Single is not available for Macros
Curious to know why you’d need it
was using it in some math-classes. converting css to values in my xml macros
typedef Single = Float then? 😅
Yup, once spoke to an Google emoloyee and the only thing they found a valid programming language was go... That engineer specifically was also a bid rude as they really only found their own opinion valid or even worthy of discussion (anything I said or the other person I was speaking to the google employee with was ignored)
I'm sure there are alot of lovely people though
got bored
you can use git submodules
look at haxe, basically a team of 3 people vs google
You know are working on a real problem when you ask your LLM about it and it locks up in an infinite reasoning loop
Lol, yeah, it can mess things up pretty quickly
I stopped asking it things with « too complex context », but it’s been very helpful for lower-commitment tasks that still require a load of code to work
ok, the css transform stack works
though, the origin should be centered
at least we can interpolate shit now
hahaa, this is awesome
Something weird in the outside part of the red border no? Looks like inner blue background is bleeding outside borders a bit
oh that's simple rounding errors due to antialiasing and nanovg trying to be smart. nothing that bothering me right now
super cool use case for ai i just had, i gave it a list of ingredients in my cupboard, and asked for it to provide me a list of dessert recipes
i literally had an idea for an app like this years ago, so damn cool
oh.. and no blog fluff piece about how some cookies changed your life!!
Recipes can be protected under copyright law if they are accompanied by “substantial literary expression.” This expression can be an explanation or detailed directions, which is likely why food and recipe bloggers often share stories and personal anecdotes alongside a recipe’s ingredients.
https://copyrightalliance.org/are-recipes-cookbooks-protected-by-copyright/
Funnily, the blog fluff piece is probably to protect against infringement...but the AI can sidestep that.
what does copyright laws have to do with cookies
I can't eat laws
whatever the reason for it existing doesn't make it less noise
Pay $10/month to access the cookies
The recipe for the cookie is available because it's profitable to provide. The profitability is protected by the copyright. The copyright is made valid by the anecdote. The AI ignores the copyright, providing the recipe from data it scraped without the original publisher(s) being able derive profitability. The progress of cookie science is disincentivized. Baking stagnates. The world of desserts crumbles and is carried away by ants.
I'll just ask the ants for the recipe
I'm sure they'll give it to me
The best recipe is probably 400 years old anyway
and everything modern is just a remix of previous genius
LLM recipes can work as long as it doesnt hallucinate battery acid into the ingredients 😜
at work we also found a pretty nice product use case for a LLM that we are currently validating. sadly cant share any details yet
These things are so useful in the right hands
How does your local setup fail vs the online ones Daz?
fair*
it's ok. local is limited by the gpu resources you can throw at it.
a big model will limit the context size available. you need to dial it in well
its very wasteful tech
how so
energy draw, a gpu will get maxed out for every peompt
depends on how you drive it. you want the most tokens per sec in/out. if you add an agentloop for tools/mcp then it multiplies
nah, defo try it
gets you a better idea and you can learn a lot about how it ticks
a cloud hosted one, see what the cost efficiency there is when I can utilise someone elses cheaper electricity access x)
hosted is heavily subsidized atm. its bonkerz, they cant sustain that pricing in the long run
take advantage for the short term, or just find a vps with sufficient hardware for your needs
vps are quite stable these days
sometime I think I should look into a virtual LLM that is distributed across machines/devices
then let people sign up their hardware and pay them from the subscriptions of the llm users
a funky p2p ai assistant? 😂
crypto backed p2p ai assistant
an actual decent use case to rally a botnet!
I think it could work by stripping LLMs and Crypto to the bone and mixing it with a little multiplayer gamedev
the game will most certainly get hacked
it cant, it's crypto, hackers will hack others to join the network to make money from foreign hardware. Awesome for the subscribers, more hardware & quicker queries
instead of mining, the machines generate partial tokens
IT'S BRILLIANT
hang on, im on the phone with ycomb
Got an seed offer for 120million for 0.2%, they want me to ship next week
🤣
You looking for work?
ok, stopping now. This might sound like satire, but dude this is how this shit rolls
free protein
given how things "look" atm i can see it being true
I worked a bit more on my Game Of Life impl - reworked the population counter and added zoom and speed controls
also set the size from 20 to 250
crypto in gaming has massive implications imo
defo something that hasn't been explored yet
crypto in gaming? hasnt that already been pulled through the gutter with metaverse and all kinds of weird mobile shit & glorified browsergame mechanics
yeah but that's just the buzzword beginnings
the actual decent thought and executions has yet to come about
also mobile games are barely games
i'd class them more as time fillers, distractions or skinned gambling apps
(turns out 1000x1000 halves the fps - I should probably use a bitset instead of an std::array<bool>)
so time wasters
not necessarily
because i think people play them because they want to "waste time"
so in a weird way, it isn't actually a "time waste"
(then again 1000x1000 is pretty huge)
What's actually fascinating with bitcoin is that it works even when nobody trust each other
trust isn't necessary when the ledger is available for everyone to read
slowly getting there. transforms are now properly transitioning. 🙂
What would be cool is something like a continuous property: if the value of something like a slider is changed while the animation is playing it should look at what point of the animation it would be at the current value (of the animation) with the new slider value and revert back to there instead of restarting the animation
This means the animation duration won't change when changing the value while the animation is playing
And it should also not jump
Sorry if it's confusing, I'm bad at translating thoughts into comprehensible English.
^^ not sure what you mean. there are no css animations in that vid. it's all transitions, meaning I just change a value and it interpolates from where ever the value currently is to the new one
what you proll mean is that rate of change is not constant over time
but that is on purpose, since CSS transition-time works like that
Yes I ment transitions
So if I saw correctly it currently restarts the animation when you change the slider's value right?
if you mean by "animation" the change over time + easing that gets applied to the attribute, then yes
now the boxes react to being clicked (:active in the CSS) and run the transition
fuck I dont interpolate the rotation correctly
fixed
Ah that's a common one 😛
yeah, shit gets normalized into unit circle when building the matrix stack, but since i decompose it again I need to reapply the actual ranges afterwards again
Yes, so what would be nice is to have some kind of property that makes it so it doesn't restart, and instead tries to smoothly go to the new value without jumping
ò.Ó, i think you are misunderstanding what is happening there. check the following vid.
you have a start and end value but you can define an easing function that can do all kinds of things, like linear, jump, bounce, etc
Yes, I understood that
But something that happens alot is the user clicking on the slider again while the slider is already transitioning
ok, lemme demostrate with a super long transition time
can you demo with a non-linear easing function?
It’s just the easing that is « easing in » at each click, but indeed that resumes where it was on subsequent clicks
So if I look closely, the "timer" of the transition gets reset on clicking
because it ramps up in speed again after every click?
right?
Or am I just braindead?
(wouldn't surprise me)
every click it applies the transition time to whatever delta there is. that is why small deltas cause slower motions while bigger deltas cause higher speed motions
It’s the same behavior as css transitions so far
Then I guess I'm braindead
im gonna look into this. the 4s animation seems to not take 4s if I disrupt it halfway
CSS transitions provide a way to control animation speed when changing CSS properties. Instead of having property changes take effect immediately, you can cause the changes in a property to take place over a period of time. For example, if you change the color of an element from white to black, usually the change is instantaneous. With CSS trans...
this seems to be different, just like I implemented
1s to complete the animation over whatever distance
these fucking people
yeah that is exactly what I felt like what was lacking
u r not braindead, for fixed intervals & reversals browser do it like you assumed, but for dynamic stuff it's what I have here.
but not only for reversing but also keeping the "velocity" when the actual new value changes
Ahhhhhh gotcha
I guess that makes me autistic then 😛
it hurts my brain, I dont wanna add this reversal stuff
it hurts to implement (i have before) but it plays a really important role
let me show an example of one game that handles transitions really well (spoiler alert: its my favorite game)
especially the large panels for wiki, beatmap listing and dashboard are cool
another good example
disruptions of the transitions feel smooth
it really feels like there is some kind of "velocity" in the transition
shows how important reversing and proper handling of disruptions are (imo)
it's actually not that complicated. I just dont want to rebuild the w3 css spec in this case
dont think osu! does that reversal logic, but rather have velocity/speed based animations instead of time-based ones
Ill just introduce a transition-mode flag where this can be toggled
If you added velocity based animations that would be extremely cool
Good question, I'll check the source tmrrw on github
Sick :D
That's cool, but you definitely don't need a whole CSS animation system to do that xD
Most definitely not
But if it could do something like that video in a reasonable way, I'd reckon that would be a good test of UI stuff overall
I see very complex transition and such all over many 3d games
In Ceramic you can use VisualTransition for that kind of stuff
The default behaviour is that if you interrupt a field transition (to create a new one) it starts from the value it was, using the new easing you provided
For more fine-grained animation, you can go one lever lower using a Tween directly and interpolate values as you want
So far it's been flexible enough to do whatever I needed
Just realized that bgfx powers Guild Wars 2
ceramic with cortex backend defo seems like a cool collab 😂
YESSSSS
I just want to create a game you know
You aren’t allowed
The tooling comes first!
I just want to trade but i'm making an elaborate application that is defo not necessary
unfortunately we're silly people
or fortunately
Tooling is awesome
If we all made games and end products, they would be bad because of no one focusing on tooling.
I'm more on the team that you need to eat your own dog food, so creating tools because you actually need them in order to make something else with it (like a game) is good. Creating tools for the sake of it or tools that you don't use yourself is fine, but afaik it never worked like this for me
i agree with you
(don't get me wrong though, I enjoy a lot the process of making tools)
tools are super fun
(except when I fight with rendering code on android heh)
but i do believe to make an exceptional tool one has to use/need it
Or you need to have a user base that needs it and be very attentive to their feedback, that can work too
it can get very hard though
yeah it's harder that way I agree
you either end up adding everyones requests and the app becomes super bloaty
or you end up with a bunch of features that only 10/10000 people use because you can't distinguish what's important and not
Oh this can be avoided even in that context, you can still care about your project architecture and sanity even if you are not the user
Given that I have been in that situation many times for stuff that is used by clients, the "importance" of a feature is somewhat linked to the size/income potential of the client xD
haha
Yeah having a single direction of the tool with a goal is way better then a tool searching for one
A tool searching for a problem is rough, and that's what you get most of the time when you build for others.
Because other people's problems are not nearly as tangible as your own.
It's easy to have a feedback loop with yourself, exceptionally hard with others.
I've been listening to a lot of the creator of elixer recently, and I love his approach to that problem.
it is possible to manage, but takes a lot of attention
He decided to just make the project as decentralized as possible and introduced meta programming so the innovation could happen in the ecosystem and he could just focus on the core language.
its similar to how i approach the server
I think that's so smart, he knew he would be too spread thin otherwise.
Oh please tell, because of the projects?
as little reliance on a central management team
limited mods
people who have channels just get to manage the channels how they want
i take on a small few
Yeah that's brilliant, it allows the community to grow much faster I think
and everyone just keeps the balance amongst themselves
And without as many problems over all
the issue with the moderation setup is that you end up following a status quo
it has to be that way, you get the perk of having a large audience
but i'm not really interested in tons of people and more interested in quality
both will have its issues
but i think the benefit of organic growth outweighs all the cons
people get to grow alongside others and i think that's cool
Longevity wise it's the winning strategy, programmers are like cats, cats are hard to herd, given some already goofy cases of people leaving and making a scene, I don't even want to imagine if this server was actually iron fisted in it's modding
cats are smarter, they understand if a place isn't to their liking they'll just leave and not cause a scene
I love this place because of that, I am a major fan of many of the projects on here and it's the only discord I use because there's no where else quite like it
if a human comes along and thinks the place isn't to their liking, they'll try to insist on changing it
Yeah good point
Also it's a huge motivator for my own project, seeing people build awesome things here too, it doesn't seem unproductive for intermittent use.
iron fist has its place, but, hard rules can't be used sparringly and i think with other approaches you have to have to be extremely rapid in your escalation
whereas in this setup, we can be a lot more selective with iron fisting
Yeah that makes sense. I see the reasoning behind rapid escalation to mitigate more issues and termoil around decisions, though it does seem like a lot of pressure, because it's often worse to second guess then do one action or the other.
there's always a price that has to be paid when you try to accelerate something
large audience at the cost of a lagging community
server grows too quickly and the community can't adapt to the size "atmosphere"
also in those kinds of servers, sharing other servers isn't allowed
here, as long as its haxe related and not advertised or spammed its pretty okay
it depends on how you do it, rather than a "don't do it"
I lack words on how awesome this is 🤯
https://www.youtube.com/watch?v=sDD6izSZInw
Ryktet har blivit verklighet – en tio minuter lång smygtitt på "Kung Fury"-uppföljaren har letat sig ut i cyberrymden! Klippet bjuder på en surrealistisk kavalkad där ingen mindre än självaste Arnold Schwarzenegger intar rollen som en cigarrökande president, beväpnad till tänderna med ett hagelgevär, redo att sätta stopp för nazis...
finally. the refactor & clean up is almost complete. one more little bug and I can look at something else
Let’s hope that movie is finally released at some point
ahhh this is soo good now
woohoo 3 days hackathon with my teams and we took a crazy idea to mvp! What a fucking endorphin rush
#dazLeadsByExample #dazNeverGiveUp

not at all. we had a rough plan.
- first day total chaos, like rats in a maze, overdozing on libs, architectures and shit
- 2nd day, sleep helped sorting ideas, hacks, prototypes.. agreement on a possible poc, fullstack punchthrough/e2e, early evening the "omg that that actually works, but holy shit this needs polish, maybe try something else?!"-moment
- 3rd day, again sleep helped, lets polish that sucker, replace/rewrite a few things, new repo, devops did their thing and late afternoon we presented our result and reviewed our initial plan. We did it and more. It's not perfect but feedback was positiv, off the charts 🙂
oh, not to forget, but we also ran quick Quake3 arena LAN matches after lunch. Happy to report that I handed out a few but long needed slaps in the form of rockets & railgun slugs
can you say what you did 👀
cant share details. But it's what I actually consider a good use-case for LLM based AIs/Agents applied to a business problem specific to our company.
Geil! That's an awesome sounding 3 day sprint :)
https://discord.com/channels/162395145352904705/1375267793053814856
and so it begins
Whats the point of advertising stuff in your nickname?
not sure what you mean "DazKind Cortex" 🙂 (or would it just be "DazCortex"?)
You know exactly what I mean "IanHaxeui"
damn, no more black coffeecups in the haxe shop
darkmode is popular 🙂
NOOOOOOO
I didn't know of the haxe shop...
Prefixing everything with 'haxe' is gonna make me go insane
So the full name is only gonna be in the github organization
Hey yall what you talkin bout
? Im specifically asking why you add your project to your nickname here in disc, you are looking for attention for it or something?
yeah, blatant self promotion like that is kinda inappropriate for this server
I wouldn't say it has anything to do with the server
u shouldnt project random shit into my simple question
(i was joking check my nickname pls)
is joke another word for scapegoat 😄
I know your nick. Im asking since I wanna form an opinion about what im observing
not enough info to judge, yet
Ohh
I'm just listing what projects I'm affiliated with - not rly looking for attention i just did it for fun
It's handy for me because I can see when if a person talks about a project, if they're related to it
But I wouldn't put it in my own name if I had a popular project because people might bother me
This is what rolling dice feels like:
https://github.com/dotnet/runtime/pull/115762
my god
at what point do you say "hey... it might actually be more efficient for me to do the work"
there are several more PRs that all resulted in a complete clusterfuck. co-pilot "fixed" tests by removing them etc
lmao
in another it fixed its tests but broken all others
copilot opened the issue??? whaaa
it also invented APIs that dont exist, documented API behavior that was plain wrong, it wrote a OS match function for osx & ios that would make newer macs be ios devices in .NET
lol
yeah, co-pilot is using all your training data to become an autonomous agent to replace SWE
i guess this is a case of giving a hammer to a baby
gun to a deaf monkey on cocaine is a better analogy
😂
sad part: they will have AI write code inside M$, putting user through even more shit
it's likely why windows 11 is crap
I see this as a good thing in the long term though
and I get it, noone wants to write COM code, but AI is not the solution to SHITTY apis
we are likely losing a full generation of software engineers that we will have to recover from
it gives a chance for reclaiming some space for those who are capable
what this is basically, is the equivalent of a microwave meal but for coding
you're not going to be able to replace a good home cooked meal
but a good home cooked meal is not a hugely complex system, its a recipe, little to no change. Imagine we would create meals like we create software
but that's the point really
a potentially poisonous explosive microwave meal.
why would you get a new "fast code" when you can build a set of re-usable "code"?
Yeah ... maybe.
quick result, compromised quality
same end result
Microwave food can be bland and non nutritious ... but at least it's mostly safe to eat.
Let's use them as a tool... not as some autonomous senior developer.
oh come on
its an analogy
it fits but it aint gonna be absolutely pin point on every detail 😂
these guys spent probably 1+ hours talking to a bot telling it to do shit for them
its the equivilent of waiting 25mins for a driver to deliver your burger that takes 20 mins to cook
Hopefully it was just done "for fun" or to give ideas .... and they won't give him merge permissions..
Hmm... all you food analogy reminding it's time to eat!
just finished my meal so its why food is on my mind 😄
but yeah, i do wonder, what would the ai give me if i told it to make a save state for my app
i recon it would try to serialise the whole thing and dump it
what's the challenge?
i see. you are using Serializer?
I use https://github.com/proletariatgames/hxinflate from time to time
what should i consider for deciding to use a serialiser?
versioning, objectcount&datasize i guess?
versioning is cool
it's a new arena for me
all past explorations of serialisation have purely been theoretcial or for "fun" rather than practical application
so it's a different kind of approach to the problem
i'm storing in a mysql db if that matters at all
cool, i guess you cant really go wrong at the current stage
yeah, versioning sounds cool on paper, but i'm perfectly happy to break and rebuild currently as well
and i'm already considering redoing the base chart setup...again 😅
well that was funny to read, I hope microsoft continues to try more AI related things: https://en.wikipedia.org/wiki/Tay_(chatbot)
That gives me an ominous feeling
I don't have a way to explain it but I'm switching from github to codeberg now 😄
can't have nice things cause a small group of people want to break your nice things x)
at least in the free model
Nobody is going to ever be responsible of software being shittier and shittier, that’s genius
'Ai wrote it, not me' 🤦♂️
Last week I tried to « vibe code » a non-critical thing at work with Claude, just to see how far it can be pushed. It’s really not there yet. It has a huge tendency to overcomplexify things, and breaking something to fix something else over and over again, in a way that it just can’t make the whole thing work properly.
i tried repeatedly telling it that its solution wasn't good and it ended up repeatedly giving me past denied solutions
it didn't take too long for it to get there as well
it doesn't do context very well
not too surprising though
i find it works best when you're like 80-90% "there" and just need a small scope thing for it
I really like deleting comments
LOL
there's a funnier side to that
"we lack documentation"
"lets make a tool that can write it for us!!"
"...hmm... I don't like all this noise my code.... delete"
its my code, i don't really need comments telling me what things do 😅
there are times where i think "should I just keep it there"
kind of like a reminder tag
oh I do this as well. I just need to search for "TODO:" and get stuff like this:
// TODO: what do we do with this callback?
// TODO: can this leak anything?
// TODO: WHAT THE FUCK! Move to config
// TODO: make sure we call this shit only once!!!
// TODO: what is this hack
i do do todo's
i never seem to actively search for them later tho 😄
i always stumble upon them by accident
I usually grab one after a break from the codebase to ease myself back in
ahh nice idea
i usually consult my list of things to do and pick something simple
the "real" todo list lol
xD
« You are absolutely right, let me redesign the whole thing »
ive had this before too, its so annoying. The only rare time i end up trying to use an llm is when I've searched online for ages and i can't find the answer to what i need, and that usually means the llm response is going to be rubbish as well 😅
I'm definitely looking forward to a world where the llm hype is over and we can move on to better things
i'm honestly starting to form a habit of skipping search engines and going straight to llms, i haven't 100% converted yet but its certainly happening of its own
the llm saves me clicks, ads, junk content and summarises things
so if i just want an easy answer to something, it can be more efficient than google
its just so imprecise i find, Im not gonna believe it until i see an actual source so its usually just an obstacle for me rather than something that helps
its like the recipe thing i mentioned not too long ago, i've searched for recipes online for years. I kind of understand the process now, whereas giving my available ingredients list to the llm, it was able to find me recipes i've never found or gave up on sifting through the search results and they worked pretty good
I think im just put off by the fact that its not actually going to give true information, its just an llm so it's just imitating plausible looking sentences in a language. That and from the few times ive used it, it hasn't been that helpful 😅
it certainly isn't precise there either, i'm not sure whether that's the source of the content or the llm, but that's fine. I don't mind optimising a good base for something
I'm also wondering if the "no ads" thing might not be the case forever
depends on where you go, i think the big guys have kind of learnt there lesson a bit
they all have monthly charges out of the gate
and they all have capped free tiers
at least afaik
Apparently even the charged tiers are a net negative on profit, at least with openai
they're all still in r&d phase right now
so they're probably taking the hit, but, they're certainly not going overboard with the freeness now
https://marketplace.visualstudio.com/items?itemName=Gruntfuggly.todo-tree
There are some extensions that automate the TODO work flow
bro I don't want that!
you cannot run from TODOs forever
they will eventually overpower with sheer numbers advantage
it ain't running, it's just a nice thing to find in the wilderness
Yooo this would be so good for me
I leave todos like a responsible person and then never follow up on them
this exists too
Lmao the urgency is based on how many O’s you have like todoooooo
Linking it to gh issues would make it worse for me, just being constantly reminded of todos like the previous one would work better
that's actually very good
the moment you realize that the duplicated events you are observing and considering a bug are actually due to event capturing & bubbling... and it's fucking awesome 😄
ok, building new controls & widgets is soo smooth now
w00t! another crazy bug fixed. The cocktail CSS selectors couldnt correctly match hierarchies involving pseudoclasses
so shit like this wasnt possible https://stackoverflow.com/a/7217397
finally reworked my rangeslider 🙂
consists of 3 controls, the background, the bar and thumb. with css you can easily animate/transition all kinds of effects
transitions & animations also emit proper events now, that way custom animations can also easily drive logic in & around controls
dazAwesome
is that the one you made for yourself?
im assuming yes considering the red-orangy inside lining of the coffee mug
yeah
incl tax + shipping
not bad
cups tend to be more expensive, but as always you never know if it's a good print till you get it
you satisfied with the print quality?
lookin forward to find out how bad my cortex hoodie turned out. till a few more days waiting
the cup is pretty good quality, like it
oh no
someone I know is using AI as a way to evaluate their idea potential
well hopefully I'm just being skeptical and they make this real 😄
xD
lol, had to checkout some hardware at work today, reviewing the images I took and now I see Haxe snuck into the image 😛
lol
How good is that laptop btw?
we run a few of these lenovo yogas, quite nice actually
SDL3+OpenGL quad moment
(also is it normal to have to call glFinish in your render function so it doesn't lag for an entire second when exiting)
Doesn’t your project have a thread now lol
this isn't lumina nor beacon
Ah
What about the fan noise?
Does it have an ARM chip? If so how well does it work for day-to-day development tasks? (I'm asking because I saw the copilot key which is for ARM-based copilot+ laptops only, but that might have changed or I'm wrong entirely lol)
There are x86 copilot laptops too
With the key?
Yeah
Mine is cursed with an NPU but doesnt have the copilot key (thank god)
it does have copilot
dang thinkpads still look as cool as they used to
are they still built like in the past?
Also WOW that thinkpad is EXPENSIVE for what it is
i5, 16gb, 256gb for almost 2k?
normal windows home
fingerprint sensor OPTIONAL
there is nothing notable about this on the surface
why is it so expensive
lol. It is for enterprise stuff. You can get them much cheaper used
Like $800
Hence why I am considering one.
Second hand thinkpad sounds like a great deal
or if you are willing to get something from 2021 you could maybe find something for free if you know the right people
The showstopper for me regarding most (windows) laptops is their heat management
Everything I tried is noisy or highly throttled as soon as you actually use it
when plugged in that is
my zephyrus G14 is actually reallyy good in terms of noise
i can't hear it 95% of the time
only when plugged in, on the dedicated gpu and gaming you can hear it
That’s the main reason I stick to macbook pro. (I don’t love Apple at all, but they really did a good job with their M1-M4 ships)
(But the price tag is very high yeah)
I hope arm windows takes off enough, so then I can use their laptops on Linux.
my school said "the laptop had to be windows"
but that is bullshit as everything is web-based and most things can be hacked to work with macos
i've quite literally paid the same as an m4 macbook pro for my laptop
which is kindof dumb imo
If i'd known beforehand I'd have defo gotten an macbook too
either dual boot the laptop or just put a windows skin and never tell anyone
but the school just wants to set the curriculum up for windows because it's just easier to manage that way
Can’t dual boot windows on apple silicon afaik
looking at the apple website
a 24GB ram, 512GB storage 15" macbook air would've prob been better for school than what i have now in terms of battery life / portability / usability
oh well
What about a windows VM?
and what is the state of wine on apple silicon?
That would definitely be progress yeah
You can use arm windows on mac with parallels. Not ideal but it works
Likely yes
I really wish overall build quality on windows pc was better, and cpu more quiet
is windows arm bad?
It’s just probably asking for trouble with various software primarily designed for x64
my 15year surface book is pretty good imo
0 fan noise
high quality hardware
but i don't think they've really refreshed the surface book line in years now
they stuck with the surface laptops i think
it's a shame really
its a good laptop
9/10 times the tech or trends that I really like get cut
Currently, I really think Apple is ahead regarding their cpu/gpu unified hardware. It’s just super pricey and very Apple-centric. I just wish this was more the default in general outside of Apple world
Discrete cards on laptops are also bringing a whole category of issues
yeah
(And they did have bad series of macbooks, their intel based touchbar macbook were awful)
But their recent series are really good hardware if you can afford it
i hear they are also throttled hard
just seems to be the standard for quiet/thin laptops
framework laptops seem pretty cool
Macbook air mostly maybe?
Because my M1 macbook pro 14 is still very fast and snappy
And if I do heavy stuff fan can occasionally turn on, but is still waaaaaay quieter than any pc I tried
Intel macs were throttled a lot however
Because it was getting very hot and noisy
i dont get the point of a very quiet pc
idrc abt some noise
besides
its good to know if your device turned on or not
fans spinning up
(for older devices) hard drive spinning
we aren't talking about PCs
noise in a PC is ok
in a laptop it is not so OK
can be very annoying because you use your laptop among other people (usually not the case with PCs)
even then its not that bad
laptop fan noise also tends to be way higher pitched
it is
have you ever sat next to someone with an rtx 4090 in their laptop? A power brick the size of an actual brick
i've never had a laptop with a dedicated GPU so idk
if you haven't then you don't understand
I can assure you it is way louder than any PC or laptop i've ever used
and my PC is pretty loud due to a half-broken water pump
being in a class with basically everyone having gaming laptops
trust me, it is distracting
I don't mind noise per say
but If I had to pick between noise and no noise
i'd pick no noise
and tbh, its less about noise for me and more about consistent noise
which is difficult for laptops in general
i can't even count the amount of times that everyone is just listening to the lesson and suddenly one laptop goes in overdrive
it's quite funny because suddenly you hear a supperr loud noise
I run a razer laptop with an integrated rtx 3080 ti. Battery life? What is that. Also you gotta speak louder, I cant hear you over the noise of my coolers and the extra fan on my desk that hauls air into it
TBS, I have an awesome noisecancelling headset and I listen to loud music all the time
oh, and concerning cooling, I dont use thermal paste in my laptop, using hair wax instead
lmao
have you got some kind of nitrogen setup
XD
unironically that does sound really cool (as a gimmick but still cool)
nitrogen gaming laptop
What started as an emergency replacement turned out to work for almost 2 years now. Can totally recommend
https://www.got2b.de/haarstyling/produkt/gel-creme-wachs/strand-matte-paste.html
I have a hard time believing that would hold my hair lmfao
it never worked for me as well
Currently using the strongest stuff Schwarzkopf has to offer
that does work well XD
but it has some shit in it for that matte look that totally allows for a good thermal bridge
im serious
not "on" but "in"
this is very cortex related
how often do you remodel the hair of your laptop? 🤣
(aka how often do you reapply the paste)
it was the outcome of me spilling a full cup of coffee over and into my running laptop. had to take all apart and clean everything, then I realized I had no more thermal paste to refit the cooler/heatpipe. It was a saturday night and I couldnt get anything from a shop for 2 days, so I yolo'ed the hair wax
see, that's the thing, that was almost 2 years ago and I never replaced it yet
can also recommend razer latops, these suckers are built like tanks
you sound like my girlfriend
yeah many people seem very happy about razer blades
😄
Nah but honestly
you putting litteral hair gel in your laptop deserves respect
I've never heard anybody mentally unstable enough to try that
You deserve a sticker atleast
In my youth I tried toothpaste and banana, so Im not exactly a noob at this
that is mad work
tooth paste = 1-2 days safe work
banana = a few hours and very tasty smell
these days most shit has sensors and emergency circuits for overheating, things blow up way less often
thats diabolical
I don't want to imagine it
I do feel like it is a double edged sword though with all the sensors
For consumer electronics that means shit turning itself off
Which does happen on the laptop of a friend I have
But also at my job which I can describe best as "heavy machinery"
The newer stuff is overall less reliable as any small difference will cause it to halt immediately
It has like tens of not hundreds if sensors
Meanwhile the older shit has like 5 sensors
@grizzled laurel
It's not animated :(
Good for you if you don't care, but I CARE 😄
i just got SDL_image working 🥲
I have been using intel PCs and Macs for a long while (laptops), they all were noisy as soon as you'd do something taxing on the cpu or gpu, unless you set them in "quiet mode" but then they are slower (throttled). Then I got a M1 macbook pro 14, and it's 99% of the time totally silent, you can run very expensive task on it and maybe sometimes the fans will start making a very small noise. It's just so much better on a daily basis. I don't want to have to always wear noise cancelling headphones. The overall build quality of the laptop is also very good (which, again, hasn't always been the case with Apple, but current series are great)
The main downside is definitely the price tag
(but worth it if you can afford it)
I think these days, apple is more price competitive than most
im never using an apple product - period
intel has always been perfect for me
anyway
MKI mitosis
Entry-level macbook airs are not that expensive for what you get for sure, but it's a bit limited on the sockets. Then if you want a better option, go with a macbook pro instead, price is rising very fast. There are for sure the highest tiers that are mostly useless for almost everyone, but Apple wouldn't miss any opportunity to get more consumer money and some consumer are willing to pay for it even if they don't need it 🤪
Good for you 👍
Ship it
I'll soon get my hands on a macbook pro 14' with M4 Pro + 48Gb of RAM. Curious to see how it performs with compilation speed, local LLMs etc...
living in the future rocks
like 10 years ago i was convinced that laptops will always stay pieces of shit compared to a cool workstation. Now Im running these specs with my razer laptop and think of workstations being pieces of shit 😄
although I stull have a threadripper setup on my bucket list. too bad AMD sucks when it comes to thunderbolt
The thing I don't like about laptop setups is the always charging battery
So now you've got a job of battery management
so desktop as a battle station will always be a thing for me
what you mean?
these days you have smart charging, my battery wont charge beyond a save threshold
Yeah charging hasn’t been an issue for me with recent macbooks
Yes, and it’s unified memory, so it can be used by both cpu and gpu, really cool stuff, especially if you run LLMs
My current setup is macbook pro 14 with M1 pro and 32Gb ram. Already very good hardware I’m happy with, but I need to get another machine to split things more clearly between work stuff and personal projects, both for practical and legal reasons going forward
hmm, better than expected
lets see how this holds up after several rounds in the washer
Do you plan to advertise your engine, so you are wearing the hoodie and stuff
Battery Health Charging
Question for you smart peeps, when you guys "read the code" is there a particular process that you follow? how do you "build the context up"?
I'm guessing its just read, reread and it happens of itself, but, just want to see
there is no concept behind any of this. I just do what my heart tells me to do
I read and establish a map in my head of the control flow & assumptions that I see
sometimes I make an excalidraw to create an actual map
concerning control flow, I segment shit into high level & low level to filter out stuff that I dont need to understand yet, otherwise I get stuck in details
Begins to answer sees the smart peep criteria 
lol, you deleted your previous answer?
no you count
i on the other hand am a certified dummy :,]
cats with a fish are the real ones
they are the smart ones
i'm just a silly hamster (with a lazer)
Yeah, for me that's pretty similar to daz'. It's mostly about having a mental map about what you are doing and the different moving parts. It's like you are zooming in an out depending on what you are doing, with more or less focus to some specific parts
holy shiiiit
I just fixed an annoying bug that was bothering me in my scroll logic for like 6 months. Was just messing with some code, scrolled past a location and was like "hmmm, ok, this is counter-intuitive, I need to reset the bounding box here..." <bug fixed/>
does your scroll logic allow scrolling thru dragging your mouse?
@tall veldt what ya doing, you type much too long
it's scary isn't it?
especially the small breaks in between
you mean touch-dragging?
by default no
If we're talking about reading codebases you haven't written yourself, it usually starts with a lot of searching. Github search sucks, so I usually download codebase, drag folder into Sublime Text, and use that to search since it just fast to open and use. All you can really do is search a public function name and see how it's used within the whole project.
Outside of that, I don't have a specific process, but I find it's heavily dependent on the codebase. A well-written project should be self-explanatory and all the pieces should be easy to understand without much context outside of the class/file. If not... welll sucks to suck. D;
I just fixed my animation glitch bug too yay
I had typed for like 10 seconds by the time you wrote this, lemme cook for a second. Ya'll are funny 
Oh yeah, quality/readability of a project you didn't write makes a HUGE difference
reminds me of someone ranting about how reading code sucks
especially code you didn't write
especially when it's shit code
and the point of the article was to say that LLM's is quite literally centered around reading code and that it is dumb for writing code in general
and how its a mess

I feel somewhat happy with the codebase for base Reflaxe, I definitely need to improve the gdscript/C++ targets a little.
But in general, I find giving things very long names with simple words >>>> short or creative names
I feel like that's the best advice when it comes to writing your own code over everything else when someone else might read it
i love verbose code
Yeee
for example, i have a pool
you can acquire and release
you may also deferRelease
so what if you want to acquire and defer a release immediately
acquireAndDeferRelease
realistically you'll only type 3 characters anyway :P
YES I have a lot of these
just stupidly verbose and descriptive names
i love them
because you get extra info
and you dont type that much extra due to autocomplete
hell yea
💯
I think I like conciseness, but never at the cost of making things more cryptic
Java-style yourOwnStupidThingProvider().getFactory().doSomeBullshit() is making me sick, but this is the extreme verbosity without actual upsides
I think the stupid patterns is what makes it dumb in java's case
Folder (package) names are also important and I really need to improve mine. Getting into a new project and trying to figure out what the HELL each one-word folder is supposed to refer to makes me go crazy. Cause you'll have a class of behavior that could technically fit in multiple categories, and things get all messy. Verbose package names are important but hard to change without breaking compatibility. T-T
I'm not of the smart peeps, but I make a lot of use of jump to definition/type definition/references, add many breakpoints as I go, and then use some test cases to get useful callstacks.
I need to see it in motion usually if it's not a simple code.
Now when I see "Provider" in a name I am having a panick attack
That I 100% agree with
I feel like I'm taking crazy pills when I open a Github page and can't even tell which folder is supposed to have the source code immediately.
This was me when I stubled upon unity project, until I figured out everything is inside Assets folder, even your lunch is there
"code is an asset" is engine programmer for "you're not allowed to crash the runtime, only I may crash the runtime"
oh i have the perfect example for this
SAAAME
i feel like i have used this example too many times already
https://github.com/ppy/osu
tell me where it draws the circle that the player hits
good luck.
oh mai gawd
enjoy :)
xD
haxe people have done pretty well at standardizing to src, just ignore openfl and kha
;)
and that lunatic ^
AHAHAHAHA
in case it makes you happy, genkit_miniaudio, the library you use in that one bell project uses source too
<3
ah just to mix things up
so just know you are using source somewhere in your project technically
making you a fellow lunatic
i keep my lunaticy contained
i doubt that
lol
I have been questioning myself if "source" wouldn't be better, but I'm too far into the src land
great question
surprise @tall veldt doesn't name his source srd
he would call it sourcaxe
reflsrc
src ftw!
dazSource
all the scrollbar math is correct now ❤️
a few more things then I can add the CSS directives. atm they emulate the browser and no extra scrollview or some other wrapper required
just a little overflow: scroll
ACTIVATE, damn you windows?
You count dummy
lots of nice responses in here - thanks!
gaaaah
I just had a legitimate use-case for haxe on nodejs
but given how fucked hxnodejs & dts2hx are these days ...
