Hi, Ingo! At 2022-06-05T10:40:43+0200, Ingo Schwarze wrote: > G. Branden Robinson wrote on Sat, Jun 04, 2022 at 08:58:14PM -0500: > > > While I have you, Deri raised a point in Savannah #62251 that I'd like > As you correctly say below, #62551.
Argh, yes. Once bug numbers get past four digits my brain likes to
start playing Mastermind with me.
> FWIW, i don't object to making "make all" and "make install" work from
> a git checkout if the tools for building groff.pdf are unavailable,
> and simply skip building and installing groff.pdf in that case. I
> expect that will cause very little complexity, not at all comparable
> to the bygone --with-doc horror show.
That's my expectation as well.
> Of course, that implies that subsequently, "make dist" will inevitably
> fail,
I think you're right, because of groff.{dvi,pdf} being in EXTRA_DIST.
If not, I'll have to force the target to fail if they're missing.
> so it won't help much for a serious developer who wants to do
> complete testing of whatever changes they are considering.
"make dist" failing will, I think, help people avoid error who
subsequently want to conduct builds from distribution archives, perhaps
due to the lower build dependency requirements. I'm thinking of *nix
distributors.
People not interested in distribution archives need not even run "make
dist" at all.
Those working on the build system (people like us) _should_ care about
distribution archives, and run "make distcheck" as a matter of course.
> Then again, maybe it helps some end-users who want to run
> groff-current without having to bother with TeX, and who are content
> with "info groff" and feel comfortable without groff.pdf.
Agreed. I think that's Deri's use case and I'd like to support it.
> P.S.
> I'm posting this here rather than in #62551 because it would really
> be badly misplaced there.
Yup, that's fine. I was merely noting the discussion's origin.
Thanks!
Regards,
Branden
signature.asc
Description: PGP signature
