On Mon, Oct 6, 2008 at 4:08 PM, Brian Paul <[EMAIL PROTECTED]> wrote: > Stephane Marchesin wrote: >> On Tue, Sep 30, 2008 at 8:34 PM, Keith Whitwell >> <[EMAIL PROTECTED]> wrote: >>> On Tue, Sep 30, 2008 at 7:21 PM, Stephane Marchesin >>> <[EMAIL PROTECTED]> wrote: >>>> Hi again, >>>> >>>> I've adapted (but didn't push it) gallivm to llvm 2.4 (svn) API >>>> changes. But then there's an obvious question about what we're doing >>>> with those API changes ? Do we settle on a given version, or do we >>>> want to adapt it as the llvm API changes ? >>> It's always been my hope that we'd be able to target stable versions >>> of llvm... at least when we're creating stable versions of our own >>> software. IE. at some future Mesa 7.6 or 8.0 stable version, we would >>> have settled on a stable llvm to go with it. >> >> Hmm ok. There are a number of side issues though, for example >> compiling clang on top of the chosen llvm version. Because clang is >> under development, there are no releases that "match" llvm releases. >> To put it clearly, you can't compile clang svn on llvm 2.2 or 2.3. >> >> Also, from discussing with the llvm developers, it seems that clang is >> still pretty buggy so they advise to use a version that's as new as >> possible (this is why I updated llvm in the first place). >> >>> However, while we're in development & on a development branch, it >>> probably makes sense to track llvm's development version. Certainly >>> I'd like to see the Cell driver make use of the llvm Cell backend, >>> which itself I understand is not fully done, so to achieve that we'd >>> want to track/contribute to the llvm svn repo. >> >> Yeah I was wondering why you were all busy working on doing a gallium >> cell backend while llvm could've done the trick. Btw there is someone >> who recently revived the cell backend in llvm so chances are it'll get >> better now. > > That's good news. I've played with the LLVM Cell code generator and > clang and found that very little of it actually worked properly. >
FWIW, you can probably blame a part of this on clang itself. Even on x86 it has issues (there's a reason it didn't get a release yet). Stephane ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Mesa3d-dev mailing list Mesa3d-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
