Re: [Libreoffice] Proofing API Performance

On Wed, 2011-11-09 at 14:04 +0200, Tor Lillqvist wrote:
> > Currently, LO seems to call the API for every word (or even letter) typed,
> > which is incredibly wasteful as grammar checking only makes sense at
> > sentence level.

        Heh :-) sounds like a beast. Also it's (no doubt) rather annoying for
the Java / LanguageTool use case I suspect.

> > I also don't really want the whole paragraph at each call;
> > just the last finished sentence.

        Right - an interesting concept.

> But what if you have an already "finished" sentence that has existed
> in the document as such for long, but then go back into it and add or
> delete one character, that changes its grammatical correctness in
> either direction? (For instance, change "it's" to "its".)

        Then we need to re-emit that sentence for grammar checking I imagine.
It seems do-able, we'd need to track cursor movement and typing a little
more cleverly (if we're not doing that already).

        Would you be interested in improving that in the core? and/or need some
code pointers for that ? presumably it'd enable heavier-duty, more
intelligent grammar checking with the same performance, which'd be
cool ?

        All the best,

                Michael.

-- 
michael.meeks@suse.com  <><, Pseudo Engineer, itinerant idiot