Re: master 79d9f3982f: Add derived-mode test for bug#24176

[Stefan Monnier]

>>> And font-lock-keywords should then be...
>>> (t (("c" 0 'font-lock-keyword-face)))?
>> I think so, yes, or maybe b+c, but it definitely should not be a+b.
> Hm...  do you see any obvious way to fix this?  We could document that
> you have to set font-lock-set-defaults to nil if you're setting
> font-lock-defaults in a derived mode, but that's not very
> satisfactory...  or we could store the value of font-lock-defaults and
> check whether it's changed (in font-lock-set-defaults)?  Hm...

I'm not completely sure either, no.
I do think we should strive to make `font-lock-add/remove-keywords` work
in a more "declarative" way, such that we can throw away
`font-lock-keywords` and recompute it at any time.


        Stefan