Bug#741573: Two menu systems

[Ian Jackson <ijackson@chiark.greenend.org.uk>]

Stuart Prescott writes ("Bug#741573: Two menu systems"):
> Ian Jackson wrote:
> > I think you are perfectly entitled to let the people who care about
> > the Debian menu take care of that testing.
> 
> As others have pointed out, that's a level a lot lower in everyone's 
> current understanding of what "should" means in the context of policy. This 
> may not be what was intended by the policy authors, but I think the average 
> maintainer reads "should" as something that *they* are supposed to do 
> unless they have a good technical reason. As Russ has pointed out, that is 
> certainly how it is presented to new maintainers in our mentors process and 
> there is an expectation there that the maintainer (not some other 3rd 
> party) is will ensure that their packages conform to the million little 
> "should"s in policy.
> 
> Policy already lists "may" as the word to use for things that are optional. 
> To me, Ian's statement above sounds a lot like a suggestion that packages 
> *may* provide trad menu files, not *should* provide.

The problem with "may" is that it suggests that there can be
situations where it is better not to provide a trad menu entry even in
cases where the policy on trad menu entries currently says that one
"should".  It implies that a judgement needs to be made between two
equally good options.

I don't imagine you're proposing changing the wording of the sections
of policy on doc-base entries, manpages, etc. to "may".

If we need to distinguish situations where we expect the maintainer to
normally do the work before uploading, and ones where we don't
necessarily, perhaps we need a new wording.

Ian.