Re: Font-lock decides function call is function declaration in C+ +

[Alan Mackenzie]

Hi, Stuart!

On Mon, Feb 12, 2007 at 10:06:22AM -0800, Stuart D. Herring wrote:
> >> It's worse than that.  Inserting a new letter that changes "foo", ...

> > is taken care of, in the commonest case.  When you type the "l" after
> > "foo", this replaces "foo" with "fool" in the cache c-found-types.

> Perhaps the cache should note how many times something is used?  Then the
> relatively common case of

> socket s;
> socket_factory sf/*=...*/;
> s.initializeFrom(sf);

> would not remove `socket' from the cache, but typo-corrections can still
> change it.

NO!!!  It isn't that sort of cache.  It's purely there to speed things
up, since without it, the various after-change functions would have to
analyse the source extensively for types for every character you type.

It's important that the cache contains most of the type identifiers.  It
doesn't much matter if one or two are missing.  What DOES matter is when
the cache has identifiers which aren't types (even if they once were).
This is what's causing the bug here.

I've got a clearer idea of where and how to remove ex-types than I did
last night.

> Davis

-- 
Alan.