Georg Baum wrote:
Am Sonntag, 9. Juli 2006 18:15 schrieb Lars Gullik Bjønnes:
No objections. Uinitialized class variables are bad.
Indeed. Nevertheless I believe that we still have a logic problem: At the
time when the coordcache is filled we should have calculated the metrics,
and the width should therefore not be uninitialized anymore. We might
even initialize the width with a big negative number as sentinel...
I see. Then there were a reason why updateMetrics() was before
fitCursor(). Two solution there:
1) re-order the logic so that fitCursor() is called after
updateMetrics() like it was before Lars changes.
2) remove the fitCursor test. Or rewrite it so that it calls
updateMetrics().
What does fitCursor means anyway?
Abdel.
