My first pro job in gaming was in porting. In my experience, while it does require knowledge in OS and rendering APIs, the platform/RHI-specific code is quite small compared to most other subsystems of a moderately-sized game.
Unless there is absolutely no abstraction, e.g.: with OpenGL being called from random parts of the game, porting is a matter of mostly gluing two interfaces together.
I have worked briefly in the game industry long ago (not in engine teams though) and ended up writing a few engines for myself after leaving, a couple quite complete.
It’s just a hobby like another. I have zero interest in writing a full game on top of it, or even publishing it.
But do you know who NEVER gives me grief about it? People I know who work at Unity or [previous employer] writing AAA engines.
My last monstrosity is a Rust one with zero dependencies other than the std-lib/OS, Metal/DX12/Vulkan, its own physics engine, spatial audio, vm for scripting and other stuff. I will probably do nothing with it other than having updated my skills in a lot of areas.
If you explore Github, you're gonna see thousands of abandoned game engines, compilers for made-up languages. And that has been happening since before the LLM era.
I used to be part of an OS messaging board in the early 2000s and almost everyone had their own OS. A dozen people or so could even run Firefox! I remember (now legend) Terry bothering us to check out LoseThos or GodOS or whatever was its name, but quite a few people had OSs that could do more than that.
Not all software needs to be commercial to be useful, even if it's just for a learning experience. I have learned a lot from those experiments, even if they're not polished.
> If you explore Github, you're gonna see thousands of abandoned game engines, compilers for made-up languages. And that has been happening since before the LLM era.
Yeah, but the people authoring those learned something.
> I used to be part of an OS messaging board in the early 2000s and almost everyone had their own OS.
Great example! I, too, once had my own toy OS, and browsed OSDev wiki nonstop. The thing is no one in the OS dev community were writing things they intended to place in front of actual users!
The difference now is that these incomplete projects:
a) Don't leave their owners any wiser than when they started, and
b) Are actually intended by their owners to be used by actual users.
> I have learned a lot from those experiments, even if they're not polished.
Would you have learned as much if you told a magic box "Make me an OS" and then slapped your name on it and uploaded it to GH?
Probably not, but sometimes I wonder if the lowered activation bump lets more people try it, even if they just vibecode the whole thing, and maybe learn a thing or two in the process? They wouldn't learn as much as they would if they did it by hand, but they would've never done it by hand in the first place - so this is better maybe? I don't know. The whole field is in a weird place right now, all previous rules of thumb might be wrong.
My gut says there are fewer people in that group than there are people who would've learned something who now decided learning something isn't worth the effort. A net loss.
> Not all software needs to be commercial to be useful, even if it's just for a learning experience. I have learned a lot from those experiments, even if they're not polished.
This is fine so long as the author is learning something (questionable) and not polluting the commons with "I made this in a weekend" vibe slop.
> This is fine so long as the author is learning something (questionable) and not polluting the commons with "I made this in a weekend" vibe slop.
Almost all of my prototypes are made this way.
"I made this in a weekend", is more often, "I made this in a day".
Like any trad-coded project, time to vibe code a backend vs frontend time is 1:N
> This is fine so long as the author is learning something (questionable) and not polluting the commons with "I made this in a weekend" vibe slop
This is exactly the same as demanding, that people who post their thoughts always post original, useful thoughts. It's just not going to happen. Making something easy will increase the total volume and the majority of that volume will be junk. It's inevitable. What is needed is a search engine for quality software. Perhaps LLMs can do that, since they are better at understanding concepts than generating them.
The banner is not just about cookies, but also about data sharing, so by accepting you increase the amount of your data that can be leaked.
These banners handle both ePrivacy consent for cookies etc, but also GDPR Art. 6(1)(a) for processing purposes (personalised ads, measurement, audience insights, precise geolocation, even device fingerprinting).
I had one of the providers recommending us that we leave the "Decline All" button out of Europe, since it's not widely prosecuted, buy recommended that we add it to California, since chances are slimmer.
Naturally in a camera meeting with a "don't tell anyone we said that" appended right before.
The marketing people in the meeting were very angry that California was "doing it to them".
It cannot do dynamic filtering [1]. Another big one I forgot about is no CNAME uncloaking. Don't get me wrong, uBOL is still a great add on, it's just not possible to do everything you could with the manifest v2 version.
uBO Lite is exactly MV3, it has most of MV3 limitations plus its own limitations comparing to Chromium. Safari just auto converts MV3 codes to Safari's own blocking engines.
Unless there is absolutely no abstraction, e.g.: with OpenGL being called from random parts of the game, porting is a matter of mostly gluing two interfaces together.
reply