Re: ancient man-db version causes false positives
On Fri, Aug 08, 2008 at 08:25:08PM +0100, Adam D. Barratt wrote:
> On Fri, 2008-08-08 at 20:59 +0200, Carsten Hey wrote:
> [...]
> > > "Adam D. Barratt" <adam@adam-barratt.org.uk> writes:
> > >
> > > > It appears to be caused by the line ".SH SKŁADNIA" ("synopsis").
> > > >
> > > > I'm assuming this is an encoding issue, as I can reproduce it on etch
> > > > but only with the locale set to "C"; in utf-8 or iso8859-1{,5} there are
> > > > no errors.
> [...]
> > One thing I still don't understand is why I couldn't reproduce this warning,
> > neither on plain etch nor in my sid pbuilder.
>
> Which locale were you using when checking on etch? As mentioned above,
> the error will only occur when the C locale is being used (e.g. by
> "LANG=C man") and is then reproducible on etch.
I use a lintian pbuilder-hook which aborts on every lintian E, W or I to build
all my packages before I upload them, so a clean sid works fine and worked fine
in the past.
The problem reproducing this under etch was mainly the missing dh_installman
since I ran man -l directly in the source.
After I downloaded and extracted the current .deb from sid I did the following:
39:carsten@taz:/tmp/d/usr/share/man/pl/man1> man -l deborphan.1 > /dev/null 2>&1 | head -n3
Reformatting deborphan.1, please wait...
deborphan.1:14: can't open `ADNIA': No such file or directory
deborphan.1:18: warning: can't find numbered character 155
40:carsten@taz:/tmp/d/usr/share/man/pl/man1> apt-get install man-db=2.5.2-2~bpo40+1
...
41:carsten@taz:/tmp/d/usr/share/man/pl/man1> man -l deborphan.1 > /dev/null 2>&1
42:carsten@taz:/tmp/d/usr/share/man/pl/man1> LC_ALL=C man -l deborphan.1 > /dev/null 2>&1
So my first guess seems to be correct, the old man-db causes the warnings.
Upgrading man-db to 2.5.2-2~bpo40+1 should fix these warnings.
Carsten
Reply to: