[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Should we standardize a way to create and maintain manpages?

Hi Nilesh,

On 2020-12-21 12:18, Nilesh Patra wrote:
> On Mon, 21 Dec, 2020, 12:22 pm Andrius Merkys, <merkys@debian.org
> <mailto:merkys@debian.org>> wrote:
>     1. Unusable for arch:any programs, as they may be cross-built.
> I do not understand this - would you mind explaining as to why would it
> affect cross building?
> Is it because of help2man marked "M-A: foreign"?
> (Not sure why that's a problem though)

During cross-build, arch:any binary packages may be built on machines
with architecture different from the target. help2man executes programs
to get their help text ('--help'). If buildd and target architectures do
not match, help2man crashes.

There was a thread on debian-devel@ about this, but I cannot locate that

>     For example, lintian warning/error reporting that
>     timestamps of d/*.1 are older than the newest upstream release? I think
>     such approach would be general enough.
> Sounds good, should we report a wishlist bug for lintian?

I would say yes, but we have to think a bit more about it. My concerns:

1. Can we trust file timestamps? I am not sure whether Git preserves
them. Also we need to think about packages in other VCSes/outside VCSes.
AFAIR, manpages may have their own timestamps in manpage text, maybe
these we can trust if createmanpages sets them.

2a. If we want to make the warning/error general enough, we need to
filter out upstream-provided manpages, which may be left not updated by
upstream's choice.

2b. If we care only about manpages generated by createmanpages (and
like) we can check only manpages directly under debian/ and probably
some other places (debian/man/ ? debian/manpages/ ? I don't know teams
policies well enough).


Reply to: