Bug 193480

SPEC looks OK but SOURCE compilations gives some warnings

# rpmlint sunifdef-*
E: sunifdef description-line-too-long Sunifdef is most useful to developers of
constantly evolving products with large
E: sunifdef-debuginfo script-without-shellbang
/usr/src/debug/sunifdef/src/ptr_vector.c
E: sunifdef-debuginfo wrong-script-end-of-line-encoding
/usr/src/debug/sunifdef/src/ptr_vector.c
E: sunifdef-debuginfo script-without-shellbang
/usr/src/debug/sunifdef/src/ptr_vector.h
E: sunifdef-debuginfo wrong-script-end-of-line-encoding
/usr/src/debug/sunifdef/src/ptr_vector.h
E: sunifdef-debuginfo script-without-shellbang
/usr/src/debug/sunifdef/src/state_utils.c
E: sunifdef-debuginfo wrong-script-end-of-line-encoding
/usr/src/debug/sunifdef/src/state_utils.c

=> Shorten your description

=> Wrong permissions on source files.
Running
find \( -name '*.c' -o -name '*.h' \) -exec chmod -x {} \;
inside of %prep fixes this.

Further issues:
* The "source-tarball" ships an autom4te.cache, i.e. it is rather carelessly
packaged. 
I recommend to remove this autom4te.cache in %prep, in advance to running
configure. I.e. you might consider to
rm -rf autom4te.cache
in %prep

* The "source-tarball" ships a precompiled sunifdef linux binary.
I recommend to remove it in %prep, i.e. consider to add
rm -rf build-bin
to %prep

* The sources contain a test suite. I recommend to add a %check section to the
spec to run it:
%check
make check

* Many of the warnings are "punned pointer" warnings. These should not be taken
lightly. They are an indication that a package contains bad code that can causem
"random" problems caused by side-effects of compiler optimization.

Thanks Ralf and Parag for your comments - I will implement all your suggestions
at the weekend.

Regarding the compilation warnings - I have already been in touch with the
author about these, and he's taking a look. I'm in the process of looking
through them myself too, to see if I can offer fixes.

(In reply to comment #3)
> Regarding the compilation warnings - I have already been in touch with the
> author about these, and he's taking a look. I'm in the process of looking
> through them myself too, to see if I can offer fixes.
Note: These warnings aren't blockers. I am willing to approve the package
without them being fixed, once the testsuite is enabled.

Updated spec and SRPM reflecting all of Ralf's comments:

Spec URL: http://physics.open.ac.uk/~ju83/sunifdef.spec
SRPM URL: http://physics.open.ac.uk/~ju83/sunifdef-1.0-2.src.rpm

Regarding the warnings - the package author has got back to me saying he's fixed
them and will make a release soon, at which point I'll update the source.

Please put the "make check" into a %check section

--- sunifdef.spec~      2006-06-03 03:30:41.000000000 +0200
+++ sunifdef.spec       2006-06-03 03:30:41.000000000 +0200
@@ -33,6 +33,8 @@
 %build
 %configure
 make %{?_smp_mflags}
+
+%check
 make check

 %install

Done. I had missed the point about %check before, sorry. I didn't actually know
%check existed. Live and learn.

Spec URL: http://physics.open.ac.uk/~ju83/sunifdef.spec
SRPM URL: http://physics.open.ac.uk/~ju83/sunifdef-1.0-3.src.rpm

APPROVED

Jonathan, would you please close this PR, once the package has been pushed to
the repos (Which AFAIS, already happened), TIA.

Package Change Request
======================
Package Name: unifdef
New Branches: F-13
Owners: brouhaha

Eric - what?

Sorry, meant to attach that to the unifdef review request, not sunifdef.  Please disregard.