Red Hat Bugzilla – Bug 182459
Build perl-Imager for FC-4
Last modified: 2007-11-30 17:11:24 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20060202 Galeon/2.0.0
Description of problem:
Build package for FC-4, Sprog requirement.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. yum install perl-Imager
Actual Results: Does not install.
Expected Results: Should install.
Test 30 fails. Investigating.
Some info here:
Don't know how much of a help this is. I notice it's the same version of t1lib
on FC-4 and FC-4.
I can't rebuild perl-Imager on devel (FC-5) from the SRPM so it seems strange
that it could ever build on devel. One (unfortunate) possibility is that since
gcc has been upgraded in between the first build of perl-Imager on devel (back
in late Sept. 2005) and now, something in the gcc version is making it fail.
gcc has also been upgraded (patched) on FC-4 in the mean time (although not a
major release, it may have backported some changes from the new gcc 4.1) and
that would maybe also explain the failure on FC-4.
I would try rebuilding perl-Imager on devel now to see if it will, in fact,
rebuild with the current gcc.
This is pure speculation, but rebuilding perl-Imager on devel should give us
I'll try Srog with out it then locally on FC-4 and get rid if it in devel to try.
I don't think perl-Imager is needed now.
(In reply to comment #5)
> I don't think perl-Imager is needed now.
You mean for Sprog? It should be built in any case, it has uses independent of
As I suspected, perl-Imager failed a rebuild on FC-5:
so this is almost certainly a compiler issue, because that's about the only
thing that has changed since the last (successful) build of perl-Imager.
Have you taken this issue upstream? Perhaps they might know what's up.
FYI, the test it fails in is t1lib related. t1lib in Extras has been
changed and rebuilt recently. If you install FC4's old t1lib-5.0.2-3
the test succeeds.
[qa@rawhide Imager-0.45]$ PERL_DL_NONLAZY=1 /usr/bin/perl
"-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')"
All tests successful.
Files=1, Tests=64, 2 wallclock secs ( 0.62 cusr + 0.20 csys = 0.82 CPU)
So, do I spec the old version?
Or speak to the t1lib packager?
Narrow it down. Verify the t1lib packaging (%changelog). t1lib 5.1.0
is from May 2005, so rather old too.
Where the test fails, it checks whether it can turn on/off the t1lib.log
file during initialisation of t1lib. It fails in line 131 where it checks
whether t1lib initialisation created t1lib.log when given the necessary
parameter. After returning, it only checks whether the log file exists.
Why exactly it doesn't exist remains to be investigated.
Here's the contents of t1lib.log when I initialise the library with
trivial C code:
(W) ScanConfigFile(): Could not open global configfile /usr/share/t1lib/t1lib.co
(W) ScanConfigFile(): Neither user's nor global Configfile has been found
(W) T1_InitLib(): Warning t1lib configuration file not found!
(W) intT1_scanFontDBase(): Font Database File FontDataBase not found!
(W) T1_InitLib(): Fatal error scanning Font Database File FontDataBase
This perl-Imager test suite is incompatible with t1lib 5.1.0.
The thing it does is calling T1_InitLib(..) consecutively without
calling T1_CloseLib() inbetweeen. t1lib doesn't re-initialise itself
when doing that.
Maybe the newer Imager 0.47 fixes this?
Doh. Should have checked that.
Success on FC-4 and Devel
At last! ;-)
Great. Closing bug. OK, now on to figuring out why Sprog fails (bug #182461)
given that it doesn't require perl-Imager anymore.