Building FireFox 1.5 SRPMs on FC4...

[Matthias Saou]

Christopher Aillon wrote :

> > Just wondering, I assume that FF 1.5 RPMs will be making their way into
> > rawhide in the next couple of days.
> > What are the chances of building their SRPMs on FC4? (x86-64 to be
> > exact)
> > Should I just give up and build the vanilla source package?
> 
> Once the s390(x) builds finish, it should hit rawhide.  I'd imagine you 
> should be able to build it on FC4 without issue, but why not try and 
> find out for sure?  :-)

I'm trying this too, and am running into major issues, related to my build
environment : I'm using mock on an x86_64 machine, in an i386 chroot for
the i386 builds.

Well, the mozilla/firefox build seems to be pretty messed up, and messy in
general, and I'm still fighting to try and *not* get the build to try and
compile x86_64 code (and 64bit asm bits) inside the i386 root!

First problem :
The configure script set --host, --target and --build to x86_64-*. This is
usually forced to the right target by %configure, which is not used here.

Second problem :
The configure "OS_TEST=`uname -m`" line causes 64bit asm to be used in
xpcom/reflect/xptcall/src/md/unix/.

Third problem :
Build starts using -m64 and other x86_64 specific flags and files inside
security/. I'm not sure where/why that starts yet, I'm still onto that one.

IMHO, when Makefiles want to detect the CPU type, they should use what was
passed as --target to the configure script, which isn't happening
everywhere here. Should I file these issues as bugs (upstream, I guess),
or are developers just not interested as long as it works for the general
case (i.e. building on the arch the binary will be built for)?

Matthias

-- 
Clean custom Red Hat Linux rpm packages : http://freshrpms.net/
Fedora Core release 4 (Stentz) - Linux kernel 2.6.14-1.1637_FC4
Load : 0.78 0.33 0.18