Nicolas Hillegeer

@nygma0451 Had fun with Blood Omen 2, because I like the universe.

I'm gonna list some overhated games. They aren't good, but they're not that bad. Others times they're just mediocre or plain okay. - Blood Omen 2 - Red Faction 2 - DmC - SoulCalibur V - Halo 4 & 5 (you read that right) - Sonic 06 - Dead Space 3 - The Order 1886 - DOA6

@RPM_PlayStation Wipeout 2048, Killzone: Mercenary, Rayman Origins But in reality there's too much: reddit.com/r/vita/comment…, also many good ports (Resogun/Geometry Wars/....)

Top 3 PS Vita games of all time, GO!

@Elgrannael I'm not even a fighting game fan, but that shadowboxing scene is amazing. Can't remember the last time I felt strong emotion when viewing a videogame trailer. Congrats to @Elgrannael for the amazing track, and the VF team for the animation. Can't wait for it to drop somewhere.

“Beyond grateful to share that my music was selected for the official trailer for this game. This is a huge milestone, and I’m excited for what’s next.” Thank you Sega for believing in my Sound . "Legacy" Rap Vocals & Lyrics by El Gran Nael

@rwdaigle @davidcrawshaw @copyconstruct Those are indeed a bane (also at DAY_JOB, where I avoid them like the plague). I hope someone knows a way to improve this, and evangelize it to the world.

@alazyleopard @davidcrawshaw @copyconstruct I’m referring to browser based e2e here - very hard to reliably test a fronted experience

end-to-end testing > unit tests, in the vibecoding era. A massive, almost entirely agent-coded refactor passed all unit and pre-merge tests but broke a critical feature. It was only caught due to my own excessive paranoia making me run end-to-end tests before the prod deploy.

@rwdaigle @davidcrawshaw @copyconstruct e2e is less brittle, because you test your contract/API (which may just be invoking a CLI if you are creating a CLI). Flakiness can be a problem, depending on what level you substitute dependencies. The answer: as much real stuff as possible (within flakiness bounds).

@davidcrawshaw @copyconstruct How do you handle the typical e2e brittleness and flakiness?

@davidcrawshaw @copyconstruct This is one of the few things about the AI era that I like. I always preferred having e2e tests for many reasons (vastly increased coverage:testlines ration, reduced maintenance, implicit docs, ...). Now, arguing for people to do integration over unit is much easier.

@copyconstruct Yes this is a major change in our programming style, most of the tests for exe.dev are end-to-end. It has required a lot of eng effort to keep our merge queue to 100 seconds with hundreds of e2e tests to run.

@royalicing @rfleury Sorry that was wrong. The right side just shows how to use it (by nesting them). The ensures that after each set of nested blocks is over, you don't have non-default OpenGL state that might make you scratch your head later.

@royalicing @rfleury It's a reference to x.com/rfleury/status… I think. The non-DeferLoop ones on the right are perhaps more complicated ones that don't have a simple "unset".

Stop writing OpenGL state machine management bugs with this one weird trick

@Rinnegatamante You mentioning vitaGL development reminds me: do you know why the RetroArch package on VitaDB is the one without a native GL (piglet) replacement. I saw it in the link (buildbot.libretro.com/nightly/playst…). Would be cool to be able to select piglet version (or have two "apps")..

A new build of Perfect Dark Vita (v.1.3) is available for download: rinnegatamante.eu/vitadb/#/info/… It restores all framebuffer effects following recent vitaGL development, fixing tons of effects in game on top of minor optimizations in the renderer pipeline.

@antirez Was a mac user from '11 to '15 (My MBP '11 was amazing). I got a MBP '15 and after a while it started randomly rebooting, losing my in progress work. Comparing logs with XNU sources said the "reason" was a corrupt network packet. Panicking? Never used Apple/macOS again.

I'm so unhappy about those kernel instabilities on macOS: "panic(cpu 11 caller 0xfffffe0054388010): cpt_mapcnt_inc: refcnt overflow: rc 0xfffffde002fabb38 old_value 2050 value 2052" Seriously? A bug that is easy to casually trigger in the virtual memory compressor implementation?

@neogoose_btw How does one prevent the autoresearch agent from just emptying the tests (or modifying the tests in other ways), if one asks it to optimize those? The examples make it look so easy, but I feel like more structure would be helpful.

This is the story how I spent $300 on autoresearch loop and significantly optimized the performance of FFF:

@teej_dv @thegenesisbl0ck Correct. I've never understood the "divide your love" take. As if love were some kind of limited quantity. Attention is. But then again, they also have the attention of their siblings, and a larger family in the future to spend happy days with.

every time you have a kid, you think there's no way i could love another kid more. and it's a little bit correct, but that's because actually you love each kid different and special because they're different and special. you see different parts of yourself and your spouse in each kid, in different scenarios and it's the best thing. plus there's nothing cuter than watching your older kids love your littler kids, which makes you love all of them even more and more

I think about this a lot. It’s so conflicting. Does the love you have for your child get multiplied when you have another, or does it get divided? I don’t ever want my first to feel less special.

@mitchellh Agree in general. Especially for a compiler. For servers case is good, too easily user-provokable expensive errors may end up being a DoS vector. Curious about the Zig compiler: does it rerun the same code paths with a flag set, or is it fully separate?

Observations from writing Go again, exacerbated by agents but not unique to them. First, its far too easy to allocate and agents (probably people too) do it too often. For example, to "undo" work on error, its enticing to keep track of the work done but that's a mistake. If an error case is rare (and they usually are), you should pessimize the error case and optimize the success case. Don't allocate unnecessarily on the happy path if its going to succeed 99+% of the time. Let the error case be slower. On error, just redo the work but do the undo step instead of the apply step. This doesn't work if the apply step had a ton of side effects but it works more often than you think. Real world example of that not in Go, but the Zig compiler: when it parses, it doesn't store any file/line/col info, because its a waste of memory when parsing succeeds most of the time. And memory is speed in modern CPUs since cache locality owns everything around us. If an error happens, Zig just reparses the file from the beginning in a slow path that does collect error information. That pattern is generally useful.

@jessfraz @jarredsumner @charliermarsh @mitsuhiko @dhh @wookash_podcast. At this point, the "Game Engineering Podcast" is a misnomer. But it's technical alright, and awesome. Nearly every episode I've listened has been a banger. Just one example: the first one with @antovsky (Stay True to Code Performance).

wheres the podcast with actual engineers doing actual engineering like give me @jarredsumner @charliermarsh @mitsuhiko @dhh people who actually code, im so tired of these "tech podcasts" & its the same people talking about the same shit & they aren't even technical!

This dude made a game engine for making PS1 games, it's such a quirky piece of hardware I honestly didn't see this happening. youtu.be/eFv2Yi0Z6E0

** Sega Genesis 3D engine Day 2 ** Added solid filled triangles. Added colors ( full 15 color buffer ) . All code in C so far apart form the buffer clear and DMA buffer sending so should be some good speedups to come yet once line / triangle routines are ported to 68k Assembly. I've done basic filling optimisations in C such as where we can write longs eg fill 8 pixels at once for interior parts of the triangle the function will utilise that. Still using 3 divisions per triangle setup , I think i will see if I can use a cheaper reciprical multiplication instead. All rotations vertices are calculated live , in 2d space at present. #SGDK #SegaGenesis #SegaMegadrive

.@runevision shared a breakdown of his GPU-friendly terrain erosion filter, designed to emulate erosion without simulation and simple to generate in chunks. The code is available: 80.lv/articles/fast-…

meshoptimizer 1.1 is out! Featuring two new major features, meshlet topology compression and opacity micromaps, as well as a few smaller features and improvements! Meshlet codec is heavily optimized for CPU decompression and output can be compressed further with Zstandard; opacity micromap rasterizer outputs hardware ready data representation and is fast and easy to integrate. Boosts and repository stars are welcome! github.com/zeux/meshoptim…

Keep in mind, these guys were not skimming through a library of mp3s - they were pulling these samples from a vast collection of vinyl records in their basement. This is art.

I switched from per mesh vertex and index buffers to a single global geometry buffer that holds everything. This is the same idea id Software used in id Tech for Doom, and you can see the vertex and index binding counts drop, which opens the road for easier GPU driven rendering.