From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i586; en-US; rv:1.4.2) Gecko/20040308 Description of problem: There are several build problems with sendmail's spec file. I am using the latest in rawhide and compiling with the latest packages in rawhide. 1) %if "%{old_setup}" == "yes" Source3: aliases %endif Using %if in Source tags is bad. This means rpmbuild will not complain if aliases is ever missing. And looking around, aliases is now missing. When old_setup yes is given, it fails to build because the aliases file is no longer in the srpm. 2) BuildRequires: util-linux is needed. If util-linux is not installed, the build fails with: /home/build/working/tmp/rpm-tmp.4227: line 1: arch: command not found + OBJDIR=obj.Linux.2.4.20-31.9smp. 3) During the install phase around line 269, there are some missing Make installs. mailstats, rmail, praliases, smrsh, and makemap all need to do a Make install-docs. Without using this target, it fails with this error messages: + /usr/lib/rpm/brp-strip-comment-note error: File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/rmail.8* error: File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/praliases.8* error: File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/mailstats.8* error: File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/makemap.8* error: File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/smrsh.8* File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/rmail.8* File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/praliases.8* File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/mailstats.8* File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/makemap.8* File not found by glob: /home/build/working/tmp/sendmail-root/usr/share/man/man8/smrsh.8* 4) Around line 379, is a series of mv statements. The files they are trying to mv is in a different directory. You get these errors: + mv /home/build/working/tmp/sendmail-root/usr/sbin/sendmail /home/build/working/tmp/sendmail-root/usr/sbin/sendmail.sendmail + mv /home/build/working/tmp/sendmail-root/usr/bin/mailq /home/build/working/tmp/sendmail-root/usr/bin/mailq.sendmail + mv /home/build/working/tmp/sendmail-root/usr/bin/newaliases /home/build/working/tmp/sendmail-root/usr/bin/newaliases.sendmail + mv /home/build/working/tmp/sendmail-root/usr/bin/rmail /home/build/working/tmp/sendmail-root/usr/bin/rmail.sendmail + mv /home/build/working/tmp/sendmail-root/usr/share/man/man1/mailq.1 /home/build/working/tmp/sendmail-root/usr/share/man/man1/mailq.sendmail.1 mv: cannot stat `/home/build/working/tmp/sendmail-root/usr/share/man/man1/mailq.1': No such file or directory 5) On older systems (RH9) there are some left over files that must be deleted: Checking for unpackaged file(s): /usr/lib/rpm/check-files /var/tmp/sendmail-rootwarning: Installed (but unpackaged) file(s) found: /usr/share/man/man1/mailq.1.gz /usr/share/man/man1/newaliases.1.gz /usr/share/man/man5/aliases.5.gz /usr/share/man/man8/sendmail.8.gz This does not show up under FC3T2. Version-Release number of selected component (if applicable): sendmail-8.13.1-2 How reproducible: Always Steps to Reproduce: 1. rpmbuild -bb sendmail.spec Actual Results: Various failure messages. Expected Results: A message saying where the rpm was written. Additional info: I will attach a patch that addresses all these issues.
Created attachment 104439 [details] Patch that addresses all 5 issues against the spec file.
1) This switch is as is. 2) This is a rpm bug. 3) Not reproducible. 4) Not reproducible. 5) RHL9 is not supported. Closing as "NOT A BUG".
>1) This switch is as is. So if people want to build with the old setup, you want it to fail because the file is not there? I can understand maybe branching around something in the install or %files phase, but not at the Source line. Why not just delete the old_setup define if its not supported? >3) Not reproducible. >4) Not reproducible. Are you testing with all the latest packages in FC3T2? I think coreutils made a subtle change in the semantics of following symlinks a month ago. This may be the source of the difference.
I will remove the switch in the near future. Yes, I have tested this with the latest packages but there were no problems in the original build process, too.