On Fri, 2006-09-29 at 17:48 -0400, David Malcolm wrote: > On Fri, 2006-09-29 at 16:31 -0300, Marcio Oliveira wrote: > > Dave, > > > > I tested your script in my FC5 and got the following results: > > > > Package tested: rpm-build-4.4.2-15.2 (FC5 original package) > > Result: If RH bug 206841 is present, expect to see a series of error > > lines complaining about missing fields > > **** Start of rpmbuild log **** > > error: Name field must be present in package: (main package) > > error: Version field must be present in package: (main package) > > error: Release field must be present in package: (main package) > > error: Summary field must be present in package: (main package) > > error: Group field must be present in package: (main package) > > error: License field must be present in package: (main package) > > > > Package tested: rpm-build-4.4.2-15.2 (patched) > > Result: success > > Wrote: /tmp/tmp.dZYmu16297/rpmbuild/SRPMS/foobarbaz-1.0-1.src.rpm > > Wrote: /tmp/tmp.dZYmu16297/rpmbuild/RPMS/noarch/foobarbaz-1.0-1.noarch.rpm > > > > After patch the rpm-build I build lots of rpm packages from tarball > > packages, and all of them are working fine in my system. > > Thanks for testing the script: looks like this is a good minimal > reproducer case for the bug. > > > > According to the tar command errors from FC5 original rpm-build > > package, tar command expects to receive "--wildcards" as a parameter > > (to accept the "*" character in "file to stract" field), and "xOvf" > > parameters plus a "-" (-xOvf) to extract the file. That is what I > > added to rpm-4.2.2/build.c file. > > > > Do you think this problem is a rpmbuild "wrong parameters problem" > > or a tar "expected parameters problem"? > > A bit of both? IIRC a lot changed in the latest "tar", and it got > fussier; does it support --wildcards in an earlier incarnation? If so, > I'm inclined to suggest that rpm-build should supply the arg so it can > work with old and new tar. > > I hope that regardless, FC6 would ship with an rpm-build/tar pair that > work together - though I'm not the maintainer of either package, just > another person who ran into the bug (hence the test) > > BTW, this is one of the tests I've written for testing Fedora in Will > Wood's "beaker" lab. We've set things up so that tests can be packaged > as RPMs, and I've got a yum repo here: > http://people.redhat.com/dmalcolm/rhts-test-repo/ > > (the RPM containing the test is the rather verbosely-named > rhts-testing108-sandbox-rpm-rpm-build-rpmbuild-of-tarball-smoketest-tests-1.1-102.noarch.rpm > - the build process takes the path of the test within our result > namespace) > > I'm gonna go and try the tests on a variety of boxes... Fails on rawhide-20060929/i386: http://tinyurl.com/n86xa Fails on rawhide-20060929/x86_64: http://tinyurl.com/po83j Passes on FC5-GOLD/i386: http://tinyurl.com/qd48s Passes on FC5-GOLD/x86_64: http://tinyurl.com/rcvgp Hope this helps Dave [snip] -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
