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

Dropping and changing some lintian checks

For the next version of Lintian, I'd like to drop or change a few checks
that have been around for a while, and I wanted to run that past everyone
here in case anyone objects or sees a reason to retain them.

I: non-standard-architecture
I: non-standard-arch-in-source-relation

These tags are issued when an architecture is mentioned that's not one of
the officially supported architectures but is still one of the
architectures recognized by Lintian.  I don't think these are ever useful.
Properly setting dependencies and architectures including the
architectures that aren't part of the main, officially-supported archive
is not a bug and Lintian shouldn't be tagging it.

Lintian will continue to warn about architectures that aren't recognized
at all.

I: several-sonames-in-same-package

This tag advises people packaging libraries to put each library in a
separate package.  This is sometimes the right thing to do and sometimes
isn't (if, for example, the SONAMEs always change in lockstep).
Furthermore, even if it would be the right thing to do in theory, it's a
bad idea to change an existing package this way without a compelling
reason.  I think this tag causes more confusion than it's worth and
propose just removing it.

W: postinst-uses-db-input

I'd like to modify this tag to only trigger if the package doesn't *also*
use debconf in a config script.  If the package is only using db_input in
postinst and not in config, that's probably a bug, but if it's using it in
both config *and* postinst, I think it's more likely that the package
maintainer knows what they're doing and Lintian should not complain.
There are various reasons to do this, such as questions that one can only
ask once the files of the package are available, or error messages that
need to be displayed during postinst.

Russ Allbery (rra@debian.org)               <http://www.eyrie.org/~eagle/>

Reply to: