On Mon, Apr 15, 2013 at 01:34:21AM +0200, Tobias Ulmer wrote: > So we have a problem that gnat buildtools sometimes fail to build > (except on my machines, grr). Parallel build issues have pretty much > been ruled out. > > The question is now whether egcc fails silently while producing > object code or gnatlink/gnatbind are doing something funny. > Since egcc/gnat1 gets called a lot during the build, I am leaning > towards the later. > > The diff below first compiles, then binds/links the buildtools with > some ls(1) thrown in between so we can see whether the object exists at > one point or not. > > Second part of this diff syncs minor changes from 4.8 to 4.7 and 4.6: > * Absolute path to strip, reduces the gnat package size by 50% > * Cpu core detection > * -j1 for the install target in 4.7, same as the rest > > I plan to commit these separate, single diff is only for testing. > > Comments? OK?
Ping. I would like to get this in before the T64 change, reduces diskspace requirements for adastrap quite a bit.