Le 06/12/2010 14:01, RGH a écrit :
So the existing situation is very close to this, except inverted: If we
notice that the cache needs updating, then we run a special routine to
update the cache (not a full buffer update), and then run the buffer
update. So if we go this way, then we have no speed gain (which was the
original hope). There's also no great simplification in the code, so far
as I can see.

I think we can run the full updateLabel thing (remember it is going to be very rare), since it is _much_ faster than the iterator-based solution.

JMarc

Reply via email to