Flags for building an rpm

[Dan Timis]

A colleague built a wine rpm a while ago.  Here is the size and date:

9922591 Apr 29 14:32 wine-20040408-1muse.athlon.rpm

If I look on the winehq site I see similar sizes, only slightly larger:

wine-20040716-1rh8winehq.athlon.rpm 11662876

Unfortunately, my colleague is incommunicado for the next couple of 
weeks, and I cannot ask him how he built the rpm.  I added a patch to 
fix the problem I reported recently in menu.c.  Then, I built the rpm 
with the command:

# rpmbuild -ba --platform athlon wine-20040408-2muse.spec

Here is what I get:

85910178 Jul 29 21:25 wine-20040408-2muse.athlon.rpm

This is significantly larger.  I looked at the contents and the problem 
seems to be mainly in /usr/lib/  All files are significantly larger.

What flags am I missing when I am building the rpm?

Thanks,

Dan Timis
Muse Research, Inc.

[Vincent Béron]

Le sam 31/07/2004 à 15:31, Dan Timis a écrit :
> A colleague built a wine rpm a while ago.  Here is the size and date:
>
> 9922591 Apr 29 14:32 wine-20040408-1muse.athlon.rpm
>
> If I look on the winehq site I see similar sizes, only slightly larger:
>
> wine-20040716-1rh8winehq.athlon.rpm 11662876

The larger size probably comes from more libraries being detected at
compilation time (like ICU), or from using a different compiler. On
which base distribution are you compiling (you and your colleague)?
> Unfortunately, my colleague is incommunicado for the next couple of 
> weeks, and I cannot ask him how he built the rpm.  I added a patch to 
> fix the problem I reported recently in menu.c.  Then, I built the rpm 
> with the command:
>
> # rpmbuild -ba --platform athlon wine-20040408-2muse.spec

My rpmbuild manpage says to use --target, while --platform is not
documented.
Else, looks good to me...
> Here is what I get:
>
> 85910178 Jul 29 21:25 wine-20040408-2muse.athlon.rpm
>
> This is significantly larger.  I looked at the contents and the problem 
> seems to be mainly in /usr/lib/  All files are significantly larger.

I've noticed it also if I build with stabs+ usage. That's why there's a
patch in the RH rpms which disables stabs+ detection (and shrinks the
archive to ~10MB).
> What flags am I missing when I am building the rpm?

It's most probably a question of stabs+ usage. For some reason, it was
disabled for the rpm your colleague did, and enabled for the one you
did.

Vincent

[Marcus Meissner]

> > Unfortunately, my colleague is incommunicado for the next couple of 
> > weeks, and I cannot ask him how he built the rpm.  I added a patch to 
> > fix the problem I reported recently in menu.c.  Then, I built the rpm 
> > with the command:
> >
> > # rpmbuild -ba --platform athlon wine-20040408-2muse.spec
>
> My rpmbuild manpage says to use --target, while --platform is not
> documented.
> Else, looks good to me...
> > Here is what I get:
> >
> > 85910178 Jul 29 21:25 wine-20040408-2muse.athlon.rpm
> >
> > This is significantly larger.  I looked at the contents and the problem 
> > seems to be mainly in /usr/lib/  All files are significantly larger.
>
> I've noticed it also if I build with stabs+ usage. That's why there's a
> patch in the RH rpms which disables stabs+ detection (and shrinks the
> archive to ~10MB).
> > What flags am I missing when I am building the rpm?
>
> It's most probably a question of stabs+ usage. For some reason, it was
> disabled for the rpm your colleague did, and enabled for the one you
> did.

Or use strip --debug on the .so files...

its just the debuginfo that makes it that large.

Ciao, Marcus