Sean Barrett

BTW, there's no magic here. Experienced coders from my peer group are unlikely to find anything of interest to this. The feedback otherwise largely seems to be that there are coders unfamiliar with the possibility that you can just sit there & continuously write code that actually gets stuff done.

@icculus@mastodon.gamedev.place 👍 A few thoughts: in stb I added a ".NO_AI" directory so it's the very first thing you see in the tree, in the hopes of increasing the chance that contributors will notice it. I didn't do an AGENTS.md, that's probably a good idea, or at least it is if an AI agent KNOWS that it's an AI/LLM, which I think they do now but will they always? What if it's "I'm a skilled programmer named Claude." FWIW, in my American English, "may" is a wishy-washy word, I'd use "must" to evoke RFC 2119.

@eschaton@mastodon.social https://noai.starlightnet.work/list.html

@Athena@chaosfem.tw (edit: whoops this was supposed to be threaded on my 'sociopaths' reply) FWIW my off-fedi followup snark: https://bsky.app/profile/nothings.bsky.social/post/3mlwge3244c27

@rygorous Vantablack Sabbath Streaminghead

@PsySal@mastodon.social I really appreciated the observation about the effect of flood-fill-visibility on the player. I haven't even played U6 specifically, but it made perfect sense to me as soon as you pointed it out as being sensible and intuitive to players.

@nicebyte@mastodon.gamedev.place There's a well-known trick for testing plane vs center+half-extent AABB which is one reason why that representation is/was popular. Was widely used for CPU frustum culling back in the day, dunno about now.

@wolf480pl@mstdn.io stb.h had a number of various kinds of allocators and I almost never use them in practice. i think this is a case where you kind of prefer to see 'through' the implementation for yourself instead of trusting a black box. with some exceptions, stb libraries tend to feel more leafy and less infrastructury

@rygorous @pervognsen I mean, "what they did to Stac" was about the fact that IIRC MS started to acquire Stac and/or Stacker, looked at Stacker's source code for due diligence, then chose not to acquire and instead released a competing product. While this COULD be totally innocent, it's not a good look, especially in the face of all the other stuff that they were doing.

@superluminal@mastodon.gamedev.place Fun fact: an invasive single-threaded profiler I wrote in 2003 built a similar call stack (actually zone stack) DAG at run-time in-process using a hash table with another zone-site based caching layer in front to make performance acceptable. It was fast enough that, until single-threaded became insufficient, the game developers who knew about it used it as an always-on feature. https://silverspaceship.com/src/iprof/ (for details, see "Implementation Notes" in the linked readme)

@Tijn Ok, I don't think this is what happened to your file, but I looked for interesting ways to corrupt a MIDI file, and I found that swapping the bottom two bits works on some of mine (but you have to be careful to only do it to the MIDI data, if you apply this corruption to whole file it fails). Rendered corrupted file:

@Tijn Rendered original file:

@tom7 indeed, one must have priorities

@tom7 Argh, my eyes!!! Should've gotten an AI to help you with the title.

@zeux@mastodon.gamedev.place @lesley@mastodon.gamedev.place @nicebyte@mastodon.gamedev.place Yeah, this is a pretty clear demonstration of that basic DoD principle that the data format you use should be whichever best fits the actual code you'll run on it.

@dougbinks I think VODs last for two weeks or two months or something, but it's also archived on YouTube (at higher resolution).

@duke_of_germany It's archived on YouTube now, if you want.

@pervognsen I dunno, I realize RAM compression is totally different from hard disk comperssion, but the way MS shit on Stac in the 90s as they pushed their monopoly--which I was there for, in my 20s--left a really bad taste in my mouth on this subject area, to the point where this sort of blog post starts to feel too hagiographic for me.

@pervognsen Unfortunately, we live in a world where the only way to get companies to support unprofitable routes is with government subsidies/contracts. We see that in the USA with trains (Amtrak) and mail-to-everyone-no-matter-how-rural (e.g. carriers using USPS as last-mile delivery).

@ubernostrum @rygorous Yeah, I don't want to be gatekeepy, but when I read Luu's examples I was like "this is more like IT work than 'programming'", and of course, no, it actually is programming, but yes, this is not the kind of programming I think Fred Brooks meant in his essay. I've elsewhere when talking about LLMs characterized this as 'the kind of programming that is programming = data structures + algorithms' that I suspect at least right now LLMs aren't great at.

@Athena@chaosfem.tw Hopefully you're proposing a serious anti-LLM RFC, but I kind of want to see "Instead of considering technical issues, we should focus on ethical issues with the decision" and expicitly include a list of pro-LLM claims that aren't allowed in discussion of the PR.

@rygorous@mastodon.gamedev.place Surely the U and V vectors are almost never completely orthogonal except in the most trivial of cases, so I guess this 'isn't useful' is based on people not caring that the direction of the perturbed normal isn't 'perfect'? I have no idea what the outcome of using "wrong" tangent/bitangent in this sense are. But yes, the "common to only want normal" is what I eventually worked out for myself is presumably a significant part of the explanation.

@zeux@mastodon.gamedev.place Why is every production render in existence throwing away bitangents and breaking bump-mapping on UV skewing, instead of throwing away NORMAL and reconstructing THAT? It's the one that's actually mathematically supposed to be perpendicular to the other two.

@zeux@mastodon.gamedev.place I mean, plausibly, there's no compact encoding of two non-orthogonal vectors (e.g. no unit vector+angle), so you just can't hit the target quality while allowing skewed UV, so skewed UV was discarded. But the thing is, even though it COULD have been a conscious choice, I don't have any evidence that it really a conscious size/quality choice, rather than just "let's do this because it's easier". Which is just kinda weird.

@zeux@mastodon.gamedev.place Huh, interesting. BTW, tangentially, I'm not clear why N = TxB would need inverse/tranpose but B=NxT wouldn't?

@zeux@mastodon.gamedev.place I don't really follow, but I think I'm missing some assumption and I'm so far from this field these days that it's not worth wasting your time further. Thanks for replying!