GNOME Bugzilla – Bug 578765
no rule to make manpages
Last modified: 2011-01-04 16:12:33 UTC
33857adc7dbf01f056a9c38d639194a4f4ba1498 removed the generated manpages, however --enable-man is still off by default, users (and buildbots) will thus experience this error: make[3]: Leaving directory `.../checkout/gnome2/gtk+/docs/reference' No rule to make target `gdk-pixbuf-csource.1', needed by `all-am'. Stop. See <http://build.gnome.org/builders/gtk%2B-RHEL5/builds/543/steps/gtk%2B%20build/logs/stdio> for example.
> however --enable-man is still off by default, users (and buildbots) will thus > experience this error: Just turn it on then ?! Users who build from git are expected to be able to handle these minor hiccups...
Created attachment 132657 [details] [review] Don't try to install manpages if not building them This patch fixes the build when not using --enable-man.
Comment on attachment 132657 [details] [review] Don't try to install manpages if not building them This is wrong. When users are building Gtk+ tarballs, we don't expect they enabling man page generation but we still want to install the bundled manpages.
Matthias, perhaps you can copy from Pango. See pango/docs/Makefile.am In short, if man rebuilding is not enabled, it creates a dummy manpage file. It also adds a dist hook to make sure such dummy man pages don't make it into a tarball. Here's the relevant bit: man_MANS = pango-querymodules.1 if ENABLE_MAN .xml.1: @XSLTPROC@ -nonet http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl $< dist-local-check-mans-enabled: @if grep "Man generation disabled" $(man_MANS) >/dev/null; then $(RM) $(man_MANS); fi else $(man_MANS): @echo Man generation disabled. Creating dummy $@. Configure with --enable-man to enable it. @echo Man generation disabled. Remove this file, configure with --enable-man, and rebuild Pango > $@ dist-local-check-mans-enabled: @echo "*** --enable-man must be used in order to make dist" @false endif
Looks like the patch from Comment #3 got applied to 2.16 branch and released in 2.16.2 (commit 663bf3db0efde02f96e34bc08846356be3240f7f by Stefan Kost, according to the 2.16.2 ChangeLog) without this bug getting touched. Wonder if Stefan didn't see this bug-report and followup discussion? Anyway, now I'm getting the user-visible breakage noted in Comment #3 whereas 2.16.1 worked fine.
the git change breaks manpage installation for distribution for example, the current 2.17 tarball is buggy too
We do what behdad proposed in commen #4, now.