On Fri, Sep 21, 2018 at 01:18:08PM -0700, Russ Allbery wrote: > Josh Triplett <j...@joshtriplett.org> writes: > > > Would there be any objection to upgrading it to a warning (not an error, > > just a warning)? > > I vaguely recall us considering doing that, but the concern was that > overrides are a mechanism to keep people using Lintian who are pretty > sensitive to unnecessary tags. So giving people a way to just make the > tag go away and not bother them about it felt a touch inconsistent with > making the unused-override tag a warning, particularly since there were > (are?) a few Lintian tags that can come and go with different versions of > the package or builds on different architectures. > > Lintian in general uses warning as the level of "we really think every > packager should know about this stuff even if you don't really like > linters or Lintian in particular," and info as the level of "okay, you > actually care about Lintian, so we're going to show you all the things we > think are reasonably fixable."
While I do think unused-override makes sense as a warning, here's an alternative that would have worked for the issue I ran into: Could Lintian's summary line mention unused overrides? The line that says: N: 1 tag overridden (1 error) could say: N: 1 tag overridden (1 error); 1 unnecessary override That would have called attention to the issue, without necessarily emitting an actual warning. Does that seem reasonable?
