Re: [ANN] Aegis 4.24.3 released
Cal Webster <[email protected]>
| Newsgroups | gmane.comp.version-control.aegis.user |
|---|---|
| Message-ID | <1311971659.9763.2673.camel@kc130ssa> |
Sorry for the delay. I had monthly status reports due... On Fri, 2011-07-29 at 09:32 +0200, Walter Franzini wrote: > Cal Webster <[email protected]> writes: > > [...] > > > Getting rpmbuild errors on the new machine right at the end of the > > build. Looks like the Aegis makefile assumes %{_prefix}/com exists. This > > is not a standard system dir so I'm not sure where the assumption comes > > from or why it builds okay on RHEL 4. Might be some changes in newer > > rpmbuild package. > > Can you try to add: > > %dir %_sharedstatedir > > to the .spec %files section? Tried it near the top, exactly as shown above: no change Tried it like this just before the "%_prefix/com/aegis" line: %attr(0755,root,bin) %dir %_sharedstatedir Still got same error... I believe this %dir directive only tells RPM that this directory is to be packaged when the RPM is built. %dir says only package the directory, not its contents, according to the RPM documentation. Since I get no errors when building on RHEL 4 as user "root" I gave this a try too but still got the same fatal error. This practice is highly discouraged anyway as it is dangerous to system stability (and security) so this isn't something I want to adopt as SOP. I was able to successfully build the SRPM as unprivileged user on a CentOS 5 i386 machine. On a Fedora 12 x86_64 machine the rpmbuild failed with the same error, though. When I check the buildroot in all the failed builds it is missing directory BUILDROOT/usr/com. Both successful builds have a full tree in BUILDROOT. I'll play around with it some more to see if I can figure out what's going on. It might have something to do with the latest rpm and rpm-build versions. RHEL/CentOS 5 was based mostly on Fedora core 6 which wasn't too far away from RHEL/CentOS 4 based on FC3. Lots of big changes in RHEL/CentOS 6, which was based mainly on Fedora 12 so it makes sense that it would fail on both. > > [...] > > > Oh yeah, and the SPEC file "BuildPrereq" tag is depreciated (causes > > build failure). The replacement is "Requires". > > Are you sure the replacement is Requires and not BuildRequires? > If I remember correctly Requires express *installation* time > dependencies. > You're right. When I got the depreciated error I skimmed over the references too fast: http://www.rpm.org/wiki/Releases/4.8.0#Deprecatedandremovedinterfaces --------------- # PreReq and BuildPreReq are now officially deprecated, with warnings at build-time # PreReq is mapped to Requires(pre,preun) at build-time --------------- BuildRequires works as expected. > [...] > > > extracting debug info > > from /home/cwebster/rpmbuild/BUILDROOT/aegis-4.24.3-1.x86_64/usr/bin/aefinish > > cpio: aegis-4.24.3/y.tab.c: Cannot stat: No such file or directory > > cpio: aegis-4.24.3/y.tab.h: Cannot stat: No such file or directory > > I'm not sure what's happening, y.tab.[ch] should be generated by > yacc|bison. Can you try to fix the BuildRequires stuff mentioned? I got these even on the old RHEL 4 machine. They weren't fatal in either case.