On Tue, 2017-01-03 at 14:15 +1000, Peter Hutterer wrote: > On Fri, Dec 30, 2016 at 01:30:51PM +0500, Mihail Konev wrote: > > As suggested, speedup partial rebuild by decreasing the number of test > > programs > > that have to be linked. > > aside from my comments in the patches themselves this looks good, thanks. > do you have any numbers on how much the speedup actually is vs the old > system? Is it worth it?
On my machine, a (warm ccache) full rebuild dropped from 65 to 58 seconds. 10% worst case ain't bad at all. If we wanted to make the build faster besides just the tests, we could do better by un-recursifying the makefiles by hand, or by just porting to something less godawful than automake. Meson is popular these days, I'm told. For that matter I'm pretty sure we could just do autoconf not automake and restrict ourselves to the portable subset of make, it's not like we're doing anything complicated. And it would free us from libtool, which Thumper's Rule tells me I'm not supposed to talk about. - ajax _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: https://lists.x.org/mailman/listinfo/xorg-devel
