Description of problem: bacula-docs package in Fedora 20. Build was stuck for 15 hours and then killed by Koji: http://koji.fedoraproject.org/koji/taskinfo?taskID=5698742 I was able to reproduce it locally with mock; all the "latex2html" commands get stuck indefinitely. An example command from the Koji build: make[1]: Entering directory `/builddir/build/BUILD/bacula-docs-5.2.13/manuals/en/developers' [...] latex2html -split 4 -local_icons -t "Developer's Guide" -long_titles 4 \ -toc_stars -contents_in_nav -init_file latex2html-init.pl \ -no_antialias -no_antialias_text \ -white -notransparent developers >tex.out 2>&1 make[1]: *** [web] Error 1 If I look into the tex.out log file I can see that the command gets stuck creating an infinite line of dots. Example: $ tail 2 tex.out | less ==> tex.out <== ;...;.. 120/213:subsection:...."Minimal Code in Console Program" for General.html ;..;... 121/213:subsection:...."GUI Interface is Difficult" for General.html ;..,....,....,...;........... ... 122/213:section:...."Bvfs API" for Bvfs_API.html ;...,.................................................. [...] The moment I stopped my mock build that was running for 3 minutes I already had an 18 mb file; 90% of it filled with dots in the last line. Version-Release number of selected component (if applicable): latex2html-2012-4.fc20 How reproducible: Always. Steps to Reproduce: 1. Try to do a scratch build of bacula-docs in rawhide Actual results: Builds gets stuck until Koji kills it. Current stuck build at the moment of writing: https://koji.fedoraproject.org/koji/taskinfo?taskID=5792896 Expected results: Build should succeed as the latest build that was done in April.
I have the same problem while building adevs package in rawhide: http://koji.fedoraproject.org/koji/taskinfo?taskID=5794017
It is regression in perl-5.18, closing as dupe of bug 978233. *** This bug has been marked as a duplicate of bug 978233 ***