upstream manpage issues

[stapelberg]

This issue documents a known limitation: some Debian packages have issues with regards to the manpages they ship.

Manpages in invalid directory

report	binary package	issue
850600	mirmon	manpages in invalid dir (man/pm/man3)
850618	controlaula	manpages in invalid dir (man/py/man1)
850619	plainbox	manpages in invalid dir (man/py/man1)
850621	dhcp-probe	manpages in invalid dir (man/cf/man5)
850623	partclone	manpages in invalid dir (man/dd/man8)
850624	sarg	manpages in invalid dir (man/man1/man1)
850625	libifd-cyberjack6	manpages in invalid dir (man/)

Manpages in wrong directory

report	binary package	issue
TODO	clang-format-3.8	manpages in wrong dir (man/man8 instead of man/man1)

Manpages with invalid names

report	binary package	issue
850641	libqwt-doc	manpages with wrong suffix (.3 inside man1)
850642	libexosip2-dev	manpages with wrong suffix (.3 inside man1)

Dangling symlinks

report	binary package	issue
696277	pyro	manpages are dangling symlinks
696277	pyro-gui	manpages are dangling symlinks
TODO	obexftp	manpages are dangling symlinks
850636	maildrop	manpages are dangling symlinks
850637	ksh	manpages are dangling symlinks
567093	ocsinventory-agent	manpages are dangling symlinks
850638	ng-cjk	manpages are dangling symlinks
850638	ng-cjk-canna	manpages are dangling symlinks
850638	ng-latin	manpages are dangling symlinks
850639	rust-lldb	manpages are dangling symlinks
850640	socks4-clients	manpages are dangling symlinks
883057	libcuse4bsd-dev	manpages are dangling symlinks
883059	pmud	manpages are dangling symlinks
883063	freebsd-utils	manpages are dangling symlinks

Quality issues

report	binary package	issue
852166	liblapack-doc-man	contains 10000 stub manpages
TODO	deja-dup	ships untranslated help output for dozens of languages
883055	postgresql	duplicate update-alternative calls

[anarcat]

awesome list. i wonder if this stuff could be automated? after all, it's "just" sending an email to the bts... ;)

[stapelberg]

Good point!

What I had in mind was working on lintian checks which catch these cases, but I’m not entirely sure yet how feasible it is to catch the different classes of errors.

In case lintian checks don’t work out, we should indeed automate the bug reporting.

One other aspect which would be good is exposing these errors to end users in an “errata” or “known issues” file, so that we don’t need to check the logs on behalf of users who are wondering about manpages not being present.