>>>>> "Martin" == Martin Vermeer <[EMAIL PROTECTED]> writes:
>> TransManager::TranslateAndInsert(char, LyXText*) (lyx) Martin> The last one is fascinating. Inside the code it says: Note that this part of the tree is collapsed. Probably TranslateAndInsert triggers lots of other things. Actually, from the following list - 19.6% TransManager::TranslateAndInsert(char, LyXText*) (lyx) - 19.6% TransManager::insert(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, LyXText*) (lyx) - 19.6% TransManager::insertVerbatim(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, LyXText*) (lyx) - 19.6% LyXText::insertChar(LCursor&, char) (lyx) - 19.6% LyXText::setCursor(LCursor&, int, int, bool, bool) (lyx) - 19.6% LyXText::setCursorIntern(LCursor&, int, int, bool, bool) (lyx) - 19.6% LCursor::setTargetX() (lyx) it is not difficult to see that all the time is spent in setTargetX, which rebreaks the paragraph. Martin> BTW about understanding this output: is it like a tree: Yes. Martin> Hmmm... but a lot less than we used to with this patch. Not in the case used by Bennett (typing in an inset). Two interesting points from the list at the end are: - 49.9% QFontMetrics::width(QChar) const (lyx) - 34.2% QPainter::drawText(int, int, QString const&, int, QPainter::TextDirection) (lyx) So we spend a _lot_ of time querying width, and if you add the text drawing (but I suspect you can't really add them), you get a lot. JMarc
