bug#13764: 24.3.50; please do not use defsubst for `font-lock-(apply-highlight|fontify-anchored-keywords)'

[Drew Adams <drew.adams@oracle.com>]

> > These are not tiny, trivial functions.
> >
> > They do not fit the criteria for `defsubst' outlined in (elisp)
> > `Inline Functions'.  And they introduce all of the disadvantages
> > listed there, especially this one, listed first:
> 
> [...]
> 
> > I see no reason why these function should not be defuns.
> 
> Presumably they are that way because of efficiency concerns.  Have you
> done benchmarking to compare the impact of these defsubsts turning into
> defuns?

Oh come on.  Why such a presumption?  Did you look at the code?

Read what you quoted above.  These do not fit the clearly
documented criteria for `defsubst'.

Do you see a reason why these should be `defsubsts'?
Have you done benchmarking to support such a supposition?

As they don't fit the Emacs criteria, and there is no code
comment that calls out why they - exceptionally - should be
defsubsts in spite of not fulfilling the criteria, a priori
they should not be defsubsts.  The burden of proof lies
with a claim that they - exceptionally - should be.

Not to mention that there is less and less need for using
defsubst.