Re: JIT stealth font-lock tuning parameters

[Stefan Monnier]

>     I'm afraid I can't.  I don't see those pauses except for the
>     ones that are reduced by syntax-ppss (which I have indeed tested and
>     confirmed that it tends to reduce their frequency).
> 
> Why don't you see them?  Is it that you have a very fast machine?

I don't think my 266Mhz PII counts as fast.

> What is different for you, such that you don't see the pauses?

I don't really know.

> Did the use of syntax-ppss make it so fast that you no longer see a pause?

That's indeed my impressions.  I used to see those pauses, but I don't
anymore now that I use syntax-ppss.  Maybe the other people will also
see their pauses disappear when font-lock-ppss is gotten rid of.
Eli, if you do suffer from those pauses, could you do
(defalias 'font-lock-ppss 'ignore) and see if they disappear ?

>     Another line of attack is a local hack (that I haven't cleaned up
>     yet) which adds a variable `quit-on-input' which forces any input
>     to cause a quit.
> 
> That is a bizarre thing to do.  If this turns out to be enough of a
> problem that we should try to solve it, I'd rather add a feature to
> parse-partial-sexp to make it check for input every so often.

But what could parse-partial-sexp do when it sees that there is
some input pending ?  How will the elisp caller be notified ?
The simplest solution I can think of is to cause a non-local exit
(i.e. a sort of quit).
That's what my quit-on-input does.  Of course, it's important to be able
to distinguish the real quit from those `quit-on-input' ones and that's
the part of my patch that still requires more work.


        Stefan