bug#21953: Eliminate warnings in the emacs-25 release branch

["John Wiegley" <jwiegley@gmail.com>]

>>>>> Dmitry Gutov <dgutov@yandex.ru> writes:

> I can get behind taking care of "correctness" warnings, but if we lose the
> obsolete functions warnings, we risk forgetting to take care of their
> causes.
> 
> And it's not like we can actually fix all instances of obsolete usages
> before the current release: a significant part of the warnings comes from
> gnus, org, cc-mode and calendar, all of which, I'm assuming, have to support
> older Emacsen.

If that's the case, I'm willing to filter those from the final count before
closing this issue. Then, in 25.2, we can focus on eliminating those as well.

> We _should_ be able to take care of CEDET warnings, but probably after their
> development migrates here.

That's good to hear; but can that happen for 25.1? CEDET is rather egregious
in the warnings category right now. If that means suppressing warnings
throughout CEDET for this release, I'm OK with that as a temporary measure.

By 25.2, anything that is built using "make" should be warning free, both C
and Emacs Lisp, using whatever reference platform we define as the litmus
test. I'll be building on that platform regularly, to report new warnings.

John