Diseased Open Source Software Community - it's about ethics in Code of Conducts

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
That one is interesting. I thought initially when I read it maybe it was just code written by people used to abusing memory overcommit (you'll see Linux/Unix apps occassionally malloc for way the fuck more memory than they actually need), which if you just ported to Windows would cause problems, since AFAIK Windows doesn't do overcommit, but it looks like it's actually using the memory.

I'm not a games programmer, but have done some scientific and industrial visualization software. I have no idea what that object format is (I'm only really familiar with wavefront), but one thing I've seen is that a lot of people just don't export normals on their objects (maybe for good reason?), and I could imagine if you're the sort of retard that doesn't think about memory at all (which is a huge percentage of modern developers), when it comes time to do shit like calculate your Gouraud normals (or whatever the kids do these days) and convert the object to a VBO (or dating myself, a display list), it would be easy for your object loader to rape the shit out the system, especially if your object format is retarded and you're not careful with allocations. I'm sure if it includes animations or trying to generate mipmaps, that only further complicates shit. Good 3D programming is kind of a dark art, and if you're a webshit developer or do street shitting GoF OO, it generally makes shit run like pure ass. Just a guess on my part, though.
If 100MB of assets are taking 6-8 GB of memory to import, there's definitely Pajeetery at hand. Either they get put into memory over and over by multiple pieces of code or there's a misplaced memory allocation inside a loop where it should've been outside it. The guy said they started having problems with a large number of asset files, so it could be the latter.
 
That one is interesting. I thought initially when I read it maybe it was just code written by people used to abusing memory overcommit (you'll see Linux/Unix apps occassionally malloc for way the fuck more memory than they actually need), which if you just ported to Windows would cause problems, since AFAIK Windows doesn't do overcommit, but it looks like it's actually using the memory.

I'm not a games programmer, but have done some scientific and industrial visualization software. I have no idea what that object format is (I'm only really familiar with wavefront), but one thing I've seen is that a lot of people just don't export normals on their objects (maybe for good reason?), and I could imagine if you're the sort of retard that doesn't think about memory at all (which is a huge percentage of modern developers), when it comes time to do shit like calculate your Gouraud normals (or whatever the kids do these days) and convert the object to a VBO (or dating myself, a display list), it would be easy for your object loader to rape the shit out the system, especially if your object format is retarded and you're not careful with allocations. I'm sure if it includes animations or trying to generate mipmaps, that only further complicates shit. Good 3D programming is kind of a dark art, and if you're a webshit developer or do street shitting GoF OO, it generally makes shit run like pure ass. Just a guess on my part, though.
I knew this for years but Godot is simply not a replacement for serious game engines and this is just one way this is made clear. During the Unity licensing drama people were advertising Godot to people who wanted to jump ship from Unity, I'm sure those were people who have never tried to develop a game in Godot. They're filling the engine with pointless bells and whistles while the core functionality is still unusable. The maintainers are asleep at the wheel, they don't know what an engine needs to have to be a serious contender.

It's telling that Godot is trying to just import everything at once, instead of building a queue. The person who wrote this has never worked on anything that needed to be performant.
 
Lol, lmao.
Apparently Godot assrapes your pc when importing assets.
i was recently decompiling a popular Godot game to look at some assets when i noticed it took a bunch of time more than it really should to import the assets. this is not an isolated issue; the engine fucking chugs at loading if you dare to work with an actual sufficiently large project.
 
Linus was more or less bullied into using the GPL. It was most definitely not his first choice, and he had concerns about it for very valid reasons, but being a pragmatist, he wanted to get the code out there and did what he had to do to make it happen.
[Citation Needed]

I read his autobiography. I don't recall any passage of the book talking about being pushed into using the GPL. I do remember passages about being raised by literal card-carrying Communists, and finding the ideas behind the GPL appealing.
 
They're filling the engine with pointless bells and whistles while the core functionality is still unusable.
To be fair, that's not a phenomenon unique to godot. Not by a damn sight. It sure seems to be the standard operating procedure with more than a few. Okay. a lot.

I'm not saying its 80% percent of FOSS, but it sure as fuck feels that way, often enough.
 
Lol, lmao.
Apparently Godot assrapes your pc when importing assets.
Note how despite the bug reporter providing a minimal reproducible example you get multiple people, including a "Godot Senior" immediately being dismissive of the issue and asking "why would you have that many assets?", with a suggested workaround from said senior to be literally "keep restarting it until it works".

I guess that sort of blame-the-user response is pretty common with open source projects but it's still frustrating to see it.
 
Knuth's O.G. TeX also comes to mind as something that's truly "complete" (though LaTeX users may disagree).
TeX is basically done which is why it's so good as an underlying; LaTeX will never be done (though it's now so ingrained into things it likely will never continue much further, either). There are simply too many LaTeX compatible weird packages on CTAN for anything else to really take off.

It's also amazing how capable raw TeX is; most of what people think as "LaTeX" is just plain TeX.
I read his autobiography. I don't recall any passage of the book talking about being pushed into using the GPL. I do remember passages about being raised by literal card-carrying Communists, and finding the ideas behind the GPL appealing.
They tried to bully him into "GPL 2.0 or later" and he adamantly refused, and refuses to consider GPL 3. He likes exactly what GPL 2 is and does.

 
They tried to bully him into "GPL 2.0 or later" and he adamantly refused, and refuses to consider GPL 3. He likes exactly what GPL 2 is and does.
Rate me 🧩 all you want, this is a different statement than the statement:
Linus was more or less bullied into using the GPL. It was most definitely not his first choice, and he had concerns about it for very valid reasons, but being a pragmatist, he wanted to get the code out there and did what he had to do to make it happen.
This statement implies that he didn't even want GPL at all, and only begrudgingly accepted GPL2, which I have seen no evidence for.

That's a different scenario than Linus picking GPL2 and sticking to it because he likes what it does and sees no advantage to later licenses.
 
Note how despite the bug reporter providing a minimal reproducible example you get multiple people, including a "Godot Senior" immediately being dismissive of the issue and asking "why would you have that many assets?", with a suggested workaround from said senior to be literally "keep restarting it until it works".

I guess that sort of blame-the-user response is pretty common with open source projects but it's still frustrating to see it.
Godot maintainers do this all the time. If you report a bug, they'll tell you you're holding it wrong, or that there's already an annoying, bothersome workaround, so they're not going to fix it. Any feature request, even if you implement it yourself, is too complex to be added, can be confusing for newcomers, is out of scope, or would require too much maintenance (except their own retarded one-off shit like volumetric shading, or an entire movie editor - those are fine).
 
Yeah was agreeing - I’ve never heard he didn’t want GPL and only seen evidence that he wants GPL 2.0 only.

BSD was in some conundrum at the time with licensing but the license itself was fine and he didn’t use it.
I have seen it speculated more than once that if BSD wasn't in the middle of legal troubles at the time, Linus Torvlads might have become a BSD kernel hacker.

I sometimes wonder how that timeline would have gone. Can you imagine the spergouts that would have been archived in mailing lists if he and Poul-Henning Kamp or Theo de Raadt had butt heads about the right way to code something?
 
Linus complains about how the GPLv3 stops Tivoization, but then complains about how other authors of the Linux kernel have sued corporations that have temporarily violated the terms of the GPLv2, because the earlier license provides no leeway whereas the newer gives users one month to fix any problems and comply.

The key lesson here is Linus complains. If the Linux kernel had never existed, Hurd would've been finished instead, and no one would've been around to take away GNU's credit.
 
Yeah was agreeing - I’ve never heard he didn’t want GPL and only seen evidence that he wants GPL 2.0 only.

BSD was in some conundrum at the time with licensing but the license itself was fine and he didn’t use it.
"BSD" was in nothing. BSDi was embroiled in a lawsuit over whether or not they'd misappropriated some code from AT&T Unix. All the PC-based forks of BSD derived from 386BSD which used precisely zero code from the original BSD, and hence was not party to the lawsuit whatsoever. However, there was so much bullshit flying around that people thought this meant BSD was encumbered and so it was avoided like the plague until after the lawsuit resolved by which time Linux was a thing. That would not, however, have stopped him from using the BSD or an MIT-style license if he wanted to. He chose to switch to the GPL because he liked it.

Ain't lawfare grand?
 
[Citation Needed]

I read his autobiography. I don't recall any passage of the book talking about being pushed into using the GPL. I do remember passages about being raised by literal card-carrying Communists, and finding the ideas behind the GPL appealing.
There was an interview in a contemporary issue of Wired from around the release of Linux 1.0 where he said he initially intended to release it into the public domain, but got very vocal feedback that he should go GPL instead, and that he basically gave in to pressure from the FSF crowd to go GNU instead of BSD or any other open source license because they were the most adamant.

Couldn't tell you the exact issue, but it was summer.
 
refuses to consider GPL 3
Linus complains about how the GPLv3 stops Tivoization, but then complains about how other authors of the Linux kernel have sued corporations that have temporarily violated the terms of the GPLv2, because the earlier license provides no leeway whereas the newer gives users one month to fix any problems and comply.
TLDR, Linus is written by many authors, everyone owning a little bit of it. It can never switch to GPLv3 unless all authors agree.
GCC could switch to GPLv3 because to contribute, you must first agree to give FSF ownership of your code via copyright assignments. I remember them sending me a physical legal letter to do that.
Linux 1.0 where he said he initially intended to release it into the public domain, but got very vocal feedback that he should go GPL instead
A few years ago, I learned that PD is not recognized within EU jurisdictions, there HAS to be an owner, making it effectively proprietary due to unclear licensing terms. A WTFPL style license would have better legal standing.
 
A few years ago, I learned that PD is not recognized within EU jurisdictions, there HAS to be an owner, making it effectively proprietary due to unclear licensing terms. A WTFPL style license would have better legal standing.
IBM mentioned this decades ago, which is why the BSD licenses exist at all.
 
What, you don't think ls and cat and grep need to be updated every five minutes to add new features?
This is a really pedantic nitpick, but those tools have evolved I noticed, to support some flags for eliminating the ansi color codes from their output for use in scripts. And the cli tools try to be cute and detect when you're passing the output to a pipeline vs when it's getting sent properly to a tty. Which is really comfy when you're aware of what might be going on but if you aren't intimately familiar with these system calls and the byte streams they use for io, you're completely mystified when you get an incomprehensible spam of square brackets and other ansi nonsense.

Yes you can potentially pull the old doctor joke. "saar it hurts when I push on my arm this way" "well stop pushing on your arm that way!"

Still, if you put out a piece of software with a couple of misconceived features, morons will quite possibly use the two together and come complaining.

It's not essential that seemingly finished tools change. But it happens anyway. And I can't really dog on the upstream authors for what seems like a fairly minor update. It might've just been a library update that conveniently adds some distinction between tty output and pipe output, which is very useful. Still, it's one more feature to be documented and maintained and propagated to downstream users.
 
Back