Paul F. Johnson wrote:
Hi,
It's not an architecture issue because it builds OK on my x86 box
but fails in mock on the very same x86 box. I thought at first it
was a missing buildreq but I can't think of anything appropriate to
add. So it's probably something to do with the mock environment
itself. Unfortunately the assertion failure is pretty unhelpful in
terms of diagnostics.
It certainly would point to a mock problem. Which architecture are you try to
get mock to build it as?
Mock on x86 can only build x86 packages (unlike x64_64).
and change the make install to
make DESTDIR=%{buildroot} install
Huh? That's what it already is (except with the macro version of
"make" used)
Just removing the possibility of a macro problem
%{__make} definitely expands to /usr/bin/make so that's not going to be
the issue.
Mono is a strange beast which (from what I can see) has little respect for
where things should be placed! ;-)
Yeah, I noticed.
I normally work it like this to eliminate problems. Build with /usr/lib set
statically, do a basic build (%configure and make DESTDIR install without any
other parameters). If that builds, move it to my x86_64 box and build. If
that's happy then I try with mock.
With each build, it's tested to ensure things aren't broken[1].
Once the builds are happy like that, then I move it to mock.
Not an option for me at the moment as I don't have access to an x86_64
box running anything later than RHEL3.
Now, if the build fails (as has happened with monodevelop on quite a few
occassions), then I start looking at where it fails and check the makefile.
In this case it's failing when running /usr/bin/mcs (whatever that is).
[1] The problem here is that I tend to have a large number of packages on my
test system, so I don't always pick up on missing deps (see my submission for
gtksourceview-sharp)
This is why I do local mock builds, which find missing deps very quickly.
Paul.
--
fedora-extras-list mailing list
fedora-extras-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-extras-list
