Sony Lands More PlayStation 4 Code Into LLVM's Clang

[phoronix]

Phoronix: Sony Lands More PlayStation 4 Code Into LLVM's Clang

Since the beginning of the year Sony has been working on landing their PlayStation 4 compiler changes back into upstream LLVM/Clang. More of that code is now hitting mainline...

http://www.phoronix.com/scan.php?page=news_item&px=Sony-PS4-Lands-More-Clang

[godlike_panos]

By quickly looking at the patch it seems the changes are PS4 specific. So this brings 2 questions:
- How will that benefit those people that don't care about PS4 (the majority of Clang users)?
- After commit, will the burden of maintenance move to LLVM/Clang devs?

[jhenke]

About your first question. It does not benefit non-PS4 users at all, like much of other code as well. You are also not benefiting from ARM changes if you only own x86 hardware. But that makes it not more or less desirable to have it, there are a lot of different use cases out there and nobody needs all of them.

About your second question. It is difficult to answer. Nobody knows how much developer time Sony will invest into maintaining this code in the future. But there is a good indication they will make sure the code is maintained and also improved. Moving the code from their own fork to upstream is meant to reduce the amount of work necessary as upstream moves along. On one hand that means upstream does have to care about not breaking stuff. On the other hand the Sony developers are contributing directly to upstream. So it cannot be clearly answered.

[Passso]

Same questions here, smells like this useless patches for 99.99999% of people will only make loose time to Clang devs instead of delivering useful features...

[elldekaa]

- How will that benefit those people that don't care about PS4 (the majority of Clang users)?
- After commit, will the burden of maintenance move to LLVM/Clang devs?

It won't, especially for the Driver/Toolchain part... But, it is the case for every not-very-popular target.
One of the main advantage of GCC is to be able to target a wide range of platforms... So we should not complain about adding some new platforms to LLVM even if Sony has obviously upstream their change to reduce their maintenance cost. Morever, LLVM development rules stricly defines that any large contribution should come with the obligation to maintain it.

[Luke_Wolf]

GPL forces someone to release code that may not even be useful: "yeaaah victory!"
Someone chooses of their own free will to release code that is mostly useful to themselves and works to get it upstreamed under a permissive license: "Boo we don't want it"

Really guys?

[gamerk2]

Define "maintain" for me? If the code works, it doesn't need to be changed unless the project itself undergoes an architectural change, in which case, sure, someone needs to port to code over.

[elldekaa]

As you said: keeping it functional (the associated functional test have to pass on buildbot). But that's the point, the architecture of LLVM is often under heavy refactoring (probably one of the reason of its simplicity/accessibility over GCC) that's why Sony have contributed to upstream. Now they will benefit from any refactoring but they still have to keep it functional.

[Guest]

GPL forces someone to release code that may not even be useful: "yeaaah victory!"
Someone chooses of their own free will to release code that is mostly useful to themselves and works to get it upstreamed under a permissive license: "Boo we don't want it"

Really guys?

Why it doesn't surprise me you're missing an obvious point? If someone releases the code under GPL (thanks God for this wonderful license!) it doesn't mean it has to be included in specific project! Those trolls..