Jean-Marc Lasgouttes a écrit :
"Abdelrazak" == Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Abdelrazak> Here is the patch. It took me a bit longer than I thought
Abdelrazak> because I did not understand the logic in the cursor in
Abdelrazak> breakParagraph(). With my former patch one call to
Abdelrazak> needsUpdateCounter on next_par was enough. With the
Abdelrazak> ParIterator version I need to test both current and next
Abdelrazak> par in order to cover all the cases. Anyway, double-check
Abdelrazak> is maybe better.
It is strange that the need for a double call is different.
Yes, I tried all kind of test based on cursor position or paragraph
emptiness but when one case did work, the other did not.
Abdelrazak> OK to apply to 1.5svn?
Yes for me. I like the patch.
OK, I will commit that and continue the cleanup from there.
Abdelrazak> Maybe it is too late for 1.4.1 but please give it a test
Abdelrazak> ride. There is nothing intrusive in there.
It is indeed late for 1.4.1, but it will definitely go in 1.4.2.
OK, then please note that there are a number of other calls to
updateCounters that can be optimized also but I did not have the time to
test them extensively. setLayout() is a good candidate for example. I
don't have a local repository for 1.4 so I'll let you manage that. OK?
BTW, do you really need the default: case in your switch? It is better
to avoid them in general...
What if there is a new kind of label that is not listed here?
With the default, at least we are sure that a full updateCounter will be
done if the label type is unknown.
Abdel.
