https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122638
--- Comment #13 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> --- > --- Comment #12 from sandra at gcc dot gnu.org --- > My main concern is lack of testing with such old makeinfo versions. I don't > want to make it harder for contributors to submit routine patches that require > documentation changes by requiring manual testing with the oldest supported > makeinfo version, and if contributors aren't checking for backward > compatibility at the time they submit patches, we're going to have the same > issues with 5.0 compatibility that we currently have with 4.7. My primary point is that right now you're worse off if you have a version of makeinfo too old to build current docs than without makeinfo at all. When it's acceptable not to build info docs when makeinfo is missing completely, it should just be as acceptable not to build them whenever it's considered too old. Currently, we just break the build in such a situation which just takes developer time for little (if any) gain. What we consider recent enough is certainly driven by the concerns you raise. Both users/developers and distro builders can easily install a recent version of texinfo if they want/need the docs. > Maybe we need a documentation buildbot that checks both the oldest > supposedly-supported version and the newest one? That would certainly help. Just relying on developers noticing (and reporting) such issues is a losing bet.
