Jean-Marc Lasgouttes wrote:
"Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> Abdelrazak Younes wrote:
Hello,
As the title says. I've been testing this patch for some days and I
didn't see any side effect. The contrary would be surprising as the
changes are pretty simple.
Will commit tomorrow if there's no objection.
Abdelrazak> done.
I still do not see what this patch gives us. here is actually no
reason when doing real painting to compute the string length, since we
are supposed to have this information in Rows (am I right?).
Well, we need the width for two things:
1) the smallcaps case needs the width information for each character
because of the font switching.
2) the underline operation needs it.
The only difference now is that the rowpainter cannot try to be clever
and avoid computing lengths in some cases.
And you reckon this is not good?
IMO any single change that minimize the frontend influence on the kernel
is good.
Metrics should do only metrics, and drawing only drawing.
Why is that? Conceptually speaking, Qt needs to know where to draw.
Retrieving the font metrics from the current painter should be cheap
because already created during the painting operation.
I think this
is a problem with the two-phase drawing (with nulpainter) approach.
Yep. I would add that this is a very difficult concept to grasp.
If I were to redesign the thing, I would just let the Frontend draw
wherever he wants. Things like line breaks etc should not be handled by
the kernel.
Abdel.
