Re: ancient man-db version causes false positives
Carsten Hey <c.hey@web.de> writes:
> Sorry for being imprecise, I thought that you are aware of this issue
> since Rhonda already reported it months ago. There is no error in
> lintian itself, the error is in the lintian setup on gluck.
>
> lintian calls "(man --warnings -l $foo > /dev/null) 2>&1" to check
> whether man reports errors in this man page. This works quite well with
> an correctly working /usr/bin/man, but on lintian.d.o an old or badly
> configured version of man is running which causes UTF-8 related warnings
> although the man page is correct.
Right, this part I knew and I thought it was already fixed. Lintian
already looks at the man-db version and suppresses some of these checks
and a bunch of warnings if it's too old.
> In pl/deborphan.1 SK£ADNIA is polish for SYNOPSIS but man thinks ADNIA
> would be a file or something like this because of the incorrectly
> working man.
This is the part that I was missing. This isn't one of the warnings that
it suppresses since I didn't know this could happen.
> The reason for this errors could be a old version of man-db or a old
> version of groff (or something completly different like non-UTF-8
> locales). A combination of man-db/bpo and groff-base/stable could also
> be the reason since lintian does not call man/stable with the option
> --warnings but it does use this option with man/bpo or man/sid and there
> were some UTF-8 related changes[1] between groff in stable and groff in
> testing.
gluck.debian.org has etch man-db, which means that lintian.d.o is not
using the --warnings flag. It sounds like this particular man page is
producing errors even without that flag.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: