Bug 189699

Note that I've finished packaging gcompris see bug 189717

I get a number of rpmlint warnings similar to this:
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos2.cc ../d_mos2.cc

rpmlint output:
E: gnucap configure-without-libdir-spec

- The configure script was not created by autoconf, and does not
  respond to a --libdir flag.

W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos2.cc ../d_mos2.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos4.cc ../d_mos4.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos6.cc ../d_mos6.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos1.cc ../d_mos1.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos7.cc ../d_mos7.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos5.cc ../d_mos5.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos3.cc ../d_mos3.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_bjt.cc ../d_bjt.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos.cc ../d_mos.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_diode.cc ../d_diode.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos_base.cc ../d_mos_base.cc
W: gnucap-debuginfo dangling-relative-symlink
/usr/src/debug/gnucap-0.34/src/O/d_mos123.cc ../d_mos123.cc

- The build system compiles against symlinks to the actual source files, which
  seems to confuse rpmbuild when it creates the debuginfo package.  You might
  try using hardlinks instead of softlinks to work around this.


MUST
====
 * Package and spec file named appropriately
 * GPL license ok, license file included
 * spec file legible and in Am. English
 * Sources match upstream:
   0d346f02a38c544fda6f8d7ec4f8b45e  gnucap-0.34.tar.gz
 * Compiles and builds in mock on:
   FC4-i386, FC4-x86_64, FC5-i386, FC5-x86_64
 * No excessive BR:
 * No locales
 * No shared libraries
 * Not relocatable
 * Does not create any directories that it should own
 * No duplicate %files
 * %clean ok
 * Contains code, not content
 * %doc ok
 * No -devel package needed
 * No .desktop file needed

QUESTIONS
=========
 * There seems to be two home pages for gnucap.  The one listed in the
   spec file at www.gnu.org lists 0.31 as the most recent version.  But
   a similarly looking page at http://www.geda.seul.org/tools/gnucap
   shows 0.34.  Why the two home pages?

MUSTFIX
=======
 * Add %{?_smp_mflags} to 'make'.  Intial tests show that it builds fine
   with this flag.

Created attachment 128286 [details]
Improbed specfile

(In reply to comment #3)
> rpmlint output:
> E: gnucap configure-without-libdir-spec
> 
> - The configure script was not created by autoconf, and does not
>   respond to a --libdir flag.
> 
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos2.cc ../d_mos2.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos4.cc ../d_mos4.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos6.cc ../d_mos6.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos1.cc ../d_mos1.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos7.cc ../d_mos7.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos5.cc ../d_mos5.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos3.cc ../d_mos3.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_bjt.cc ../d_bjt.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos.cc ../d_mos.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_diode.cc ../d_diode.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos_base.cc ../d_mos_base.cc
> W: gnucap-debuginfo dangling-relative-symlink
> /usr/src/debug/gnucap-0.34/src/O/d_mos123.cc ../d_mos123.cc
> 
> - The build system compiles against symlinks to the actual source files,
which
>   seems to confuse rpmbuild when it creates the debuginfo package.  You might

>   try using hardlinks instead of softlinks to work around this.
> 

I noticed this too, this also came up during a review  by me of kismet which
generates the same warnings. We (I and the kismet packager) concider this an
rpmbuild bug and have bugzilla'd it, see:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=189928


> QUESTIONS
> =========
>  * There seems to be two home pages for gnucap.  The one listed in the
>    spec file at www.gnu.org lists 0.31 as the most recent version.  But
>    a similarly looking page at http://www.geda.seul.org/tools/gnucap
>    shows 0.34.  Why the two home pages?
> 

I noticed this too, it seems that gnucap @ gnu is dead, and that the geda
project has taken it over.


> MUSTFIX
> =======
>  * Add %{?_smp_mflags} to 'make'.  Intial tests show that it builds fine
>    with this flag.
> 

Hmm, unfortunatly I don't remember if I left that out on purpose or by
accident,  since there is no comment about it (there is in gcompris.spec) I'll
assume it was left out by accident and have put it in, this builds fine for me
too.

As usual during CET office hours I'm behind a machine which doesn't have upload
access to my homepage, so I've attached the modified specfile.

(In reply to comment #4)

> > W: gnucap-debuginfo dangling-relative-symlink
> > /usr/src/debug/gnucap-0.34/src/O/d_mos123.cc ../d_mos123.cc
> > 
> > - The build system compiles against symlinks to the actual source files,
> which
> >   seems to confuse rpmbuild when it creates the debuginfo package.  You might
> 
> >   try using hardlinks instead of softlinks to work around this.
> > 
> 
> I noticed this too, this also came up during a review  by me of kismet which
> generates the same warnings. We (I and the kismet packager) concider this an
> rpmbuild bug and have bugzilla'd it, see:
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=189928

I agree that it's a bug in rpmbuild, but you still end up producing an almost
uselss debuginfo package.  I found a simple workaround.  Add the following two
lines to %prep to use hardlinks instead of softlinks:

%{__sed} -i 's!ln -s!ln!g' src/Make1
%{__sed} -i 's!ln -s!ln!g' src/Make.aux

> > QUESTIONS
> > =========
> >  * There seems to be two home pages for gnucap.  The one listed in the
> >    spec file at www.gnu.org lists 0.31 as the most recent version.  But
> >    a similarly looking page at http://www.geda.seul.org/tools/gnucap
> >    shows 0.34.  Why the two home pages?
> > 
> 
> I noticed this too, it seems that gnucap @ gnu is dead, and that the geda
> project has taken it over.

In that case the URL tag should point to the Geda project page, since the gnu
project page hasn't been updated since version 0.31, almost 4 years ago.  Add a
 comment with a pointer to the gnu.org page if you feel it's necessary.

Created attachment 128312 [details]
improved specfile

(In reply to comment #5)

> I agree that it's a bug in rpmbuild, but you still end up producing an almost

> useless debuginfo package.  I found a simple workaround.  Add the following
two
> lines to %prep to use hardlinks instead of softlinks:
> 
> %{__sed} -i 's!ln -s!ln!g' src/Make1
> %{__sed} -i 's!ln -s!ln!g' src/Make.aux
> 

Ok, I agree that this is a good workaround for now, added.

> In that case the URL tag should point to the Geda project page, since the gnu

> project page hasn't been updated since version 0.31, almost 4 years ago.

My bad, I though the gnu page went all the way to 0.34 and the only thing the
geda page had extra were the unstable snapshots, fixed.

Looks much better now.  All MUST items fixed.  I'll trust you that that the
package works since I'm not familiar enough with how to use it.

Now I can start the long process of reviewing gcompris.  :)

APPROVED

Imported and build, Thanks!