bug#24870: 26.0.50; parse-partial-sexp ignores comment-end

[Alan Mackenzie]

Hello, Andreas.

On Thu, Dec 15, 2016 at 09:18:01AM +0100, Andreas Röhler wrote:


> On 14.12.2016 20:56, Alan Mackenzie wrote:

> > We're talking about 9dcf5998935c8aaa846d7585b81f0dcfe1935b3d from Sun
> > Mar 20 13:19:48 2016 +0000, still?

> > The idea is that in a (parse-partial-sexp from to), the end position
> > might be in the middle of a two character comment marker, such as "/*".
> > Before this change, it was impossible successfully to use the result of
> > that operation as the old state for continuing parse-partial-sexp from
> > that position, since it did not contain enough info to see it was in a
> > comment after passing the "*"

> > The change 9dcf599 added an extra element onto the parse state which was
> > non-nil when we end up after a "/", etc.



> Hi Alan,

> sounds like a classical mistake for me.

Quite possibly.

> You commented lately on the effect of narrowing and how simply to 
> respect its results. Nothing further to say here.

I don't see what your meaning is here, but never mind.

> OTOH: do you have a use-case, a bug, which propelled the amendment?

Yes.  It was quite a few years ago, but a bug in CC Mode was caused by
parse-partial-sexp terminating at a critical place, and the next
invocation of parse-partial-sexp thus going wrong.  I programmed round it
awkwardly at the time.

Also syntax-ppss would be falling into the trap quite a lot, I think.  I
don't think it checked specially for the critical case.  Now it doesn't
have to bother - at least, it won't as soon as Noam has corrected the
current bug.  ;-)

> Thanks,

> Andreas

-- 
Alan Mackenzie (Nuremberg, Germany).