Re: gnustep-base mysterious FTBFS on powerpcspe

[Yavor Doganov <yavor@gnu.org>]

On Mon, 19 Feb 2018 15:08:58 +0200,
John Paul Adrian Glaubitz wrote:
> On 02/19/2018 01:49 PM, Yavor Doganov wrote:
> > gnustep-base/1.25.1-2 failed to build on powerpcspe [*] but
> > unfortunately I cannot understand the reason.
> 
> This sometimes happens on the machines pathfinder and atlantis but
> not cb30. For some reason the kernel used on these two machines has
> out-of-memory issues despite the fact those machines are equipped
> with 8 GiB RAM each.

Thanks for the info, sounds like a bug that is hard to identify and fix.

> Will give the package back. It usually works the next time.

Unfortunately it failed again (on atlantis):

| dpkg-deb: building package 'gnustep-base-runtime-dbgsym' in '../gnustep-base-runtime-dbgsym_1.25.1-2_powerpcspe.deb'.
| dpkg-deb: error: <compress> from tar -cf subprocess was killed by signal (Killed)
| dh_builddeb: dpkg-deb --build debian/.debhelper/libgnustep-base1.25/dbgsym-root .. returned exit code 2
| dpkg-deb: building package 'libgnustep-base1.25' in '../libgnustep-base1.25_1.25.1-2_powerpcspe.deb'.
| dh_builddeb: Aborting due to earlier error

I'm quite surprised that gnustep-base is hit by this OOM issue, it's a
dwarf package compared to many other beasts like libreoffice, webkit,
gcc, etc.  Could it be a powerpcspe-specific problem introduced in
recent binutils?