On 08/07/2011 02:25 AM, Jose Fonseca wrote: > ----- Original Message ----- >> On Sat, Aug 6, 2011 at 5:07 AM, Chad Versace <c...@chad-versace.us> >> wrote: >>> On 08/05/2011 05:41 AM, Dan Nicholson wrote: >>>> I haven't >>>> seen any Android.mk files show up in freetype or expat or anything >>>> like that. In the same way, mesa doesn't carry a debian folder >>>> even >>>> though that's how debian and ubuntu build mesa. Certainly if there >>>> are >>>> fixes to the existing build infrastructure that help get mesa >>>> built on >>>> android, that should be done, but I don't see why we should carry >>>> the >>>> android build bits in upstream mesa. >>>> >>>> -- >>>> Dan >>> >>> Dan, you make a strong point. There's no sense in cluttering Mesa >>> with >>> additional makefiles, at least not now. Other projects do not do >>> that for >>> Android, and Mesa doesn't even do it for beloved Debian. The best >>> approach would >>> be to maintain the Android makefiles in a separate branch, perhaps >>> even in a >>> personal repo. >> I do not have a strong motivation to upstream Android support because >> >> - it requires changes to Android framework >> - it introduces yet another build system to Mesa >> >> I said this a few months back on mesa-user or -dev when asked. >> >> But generally, I think the not cluttering Mesa with another build >> system is debatable. I know I conflict with myself here, but if >> someone ports Mesa to arguably the most widely used OS and would like >> to upstream it, should we shut it down because the OS _requires_ its >> own build system? Incidentally, gstreamer has Android.mk upstream. > > Personally, I wouldn't object having the Android.mk on master branch. IMO > code should be useful, above being beautiful and clean. As long it is > maintained, serves a purpose, and does not impair the current stuff, then > it's fine by me FWIW. > > Jose
Having considered this more, I see two large benefits from having the Android.mk's on the master branch. 1) It would relieve Android Mesa maintainers (including me) from the need to periodically rebase the Android branch atop master. 2) More importantly, it would enable anyone working on Android Mesa to simply pull from master and to be able to build and test with all the latest bits. I think (2) is a strong point for keeping the Android.mk's on master. Since the Android port will now be undergoing active development, I believe the inability to simply pull from master and then build and test with the latest bits is a significant, and unnecessary, hindrance to Android GL development. Since Jose is ok with presence of the Android.mk's and no one has posted any vehement objections, I will begin moving forward with this. (This patch series is annoyingly long, and some patches have bugs, so, as time permits, I will repost the series as several smaller, more digestable series of 3 to 4 patches each. I'm at SIGGRAPH this week, so expect the emails to be intermittent.) -- Chad Versace c...@chad-versace.us
signature.asc
Description: OpenPGP digital signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
