Hacker News new | ask | show | jobs
by yincrash 6 hours ago
It's interesting to read that an entire video streaming rework was done with AI, but likely won't be upstreamed because of the upstream policy on AI gen code. I wonder if that will ever be re-done using the plan GE outlined in the release notes.
3 comments

Also really excited about this because there are several games that get kind of borked because they get stuck on an intro video before you get to the game menu without doing any winetricks.

> With all of that work done, I am happy to say all of the games listed above now have functional video playback with NO winetricks needed and NO dll overrides needed. No quartz,no dshow, no amstream, no lavfilters, no klite, no rsx3d, no wmp9, no wmp11, etc. -- All the functionality previously needed from those overrides is now patched directly into wine for the listed games that needed them, and the protonfixes that were previously added have now been removed since they are no longer needed.

Yeah the big left 4 dead clone (forgot the name) had video playback issues for me.
It might come to be that maintainers have to begin accepting llm-authored or llm-assisted contributions just to maintain control of the project. Otherwise users will gravitate towards forks that offer the functionality they want.
They very specifically do not accept AI contributions because there is no way to tell if it's just regurgitating parts of the various Windows source code leaks from over the years ad-hoc, which would be a very costly mistake to make if Microsoft were feeling litigious.
The remedy for unintentional infringement is generally to remove the infringing code and cease distribution. That used to be a serious issue when rewriting the offending code might take years. But these days? Rewriting any offending code is a matter of specifying the interfaces and setting Claude / Codex to work. Risk of incorporating derived code might go up with accepting LLM submissions, but cost of recovering from them seems to have dropped accordingly, at least on the technical side.
Tech companies shouldn’t be able have it both ways and say that copyright doesn’t apply to LLM-generated code only when it benefits them.
They likely won't, once precedent setting judgements start coming out
And yet, they likely will be able to. The law doesn't have much hold on the rich.
That would be something Microsoft would have to prove in court and not that the AI came up with a similar approach on its own. ReactOS never got sued despite its similarity to Window's code.

Also not all Wine code is related to reverse engineering.

On the other hand, it'd be absolutely fascinating to see how that'd play out. The ramifications could be huge.
More reason to ditch C, C++, and C#!
The language wine is written in is irrelevant, the source is still tainted.
Or it might come to be that rejecting LLM-authored or LLM-assisted contributions becomes a badge of quality, and users gravitate to them to avoid buggy, inconsistent, or non-performant versions of the same software.
There's no need to guess, the reasoning is clearly laid out:

“Don't use an LLM tool to generate code. There's no guarantee that the training material of that LLM respects our Clean Room Guidelines, or that its output is compatible with the LGPL.”

--https://gitlab.winehq.org/wine/wine/-/wikis/Clean-Room-Guide...

The allure of being just good enough where the bling and hype of features outshine hidden bugs may win out socially.
Doesn't wine have various rules to remain a white-room implementation?

Not sure using LLMs which have possibly been trained on leaked Windows sources would be compatible with that. But that's just speculation, I wonder if LLMs possibly using leaked sources for training has been looked into. (probably legally difficult as the investigator would have to access the leaked sources too...)

I've implemented several missing features and fixed several compatibility issues in wine using LLMs. It works very well for the use case.