On Sun, Oct 22, 2017 at 04:13:50PM -0400, Sean McGovern wrote: > On Oct 22, 2017 12:40, "Diego Biurrun" <[email protected]> wrote: > On Wed, Oct 18, 2017 at 03:22:15AM +0200, Diego Biurrun wrote: > > This reverts commit 4822ee3ca620a92cd2b0a9a03ea9e34288192c79. > > AIX is a fringe oddity. Do not clutter the main codepath with AIX > > workarounds. > > --- > > configure | 4 +--- > > 1 file changed, 1 insertion(+), 3 deletions(-) > > ping > > Given the issues we've seen all the way back in 2013 [1] -- especially the > grief we had with known-broken functions in the math libraries and standard > I/O, maybe we should stop supporting this platform altogether. I'm somewhat > suspicious of the green AIX FATE nodes but don't currently have the cycle > to investigate it. > [1] https://lists.libav.org/pipermail/libav-devel/2013-July/048369.html > (and surrounding patches with AIX in the title) > > This would probably only involve removing compat/aix. > > Opinions? Bikeshed colours?
AIX support is less intrusive than other legacy proprietary Unix flavors, but it is certainly completely pointless in the year 2017. Quite certainly AIX is still used to run certain legacy banking applications, but nothing multimedia-related. If you are into retrocomputing, you also run retro software, not just retro hardware, so even that niche usage does not require the support in our codebase. I wonder when the AIX FATE instances turned green. We had a pernicious issue with AIX stubbornly printing different values than other platforms. I blank out on the exact details unfortunately. So I certainly won't fight to keep AIX support. Upcoming is an RFC patch that removes support for it to show what the actual changes are. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
