mwolff added a comment.
When I open your fake file in kwrite, then raise the line length limit and wait for the file to be rendered (which takes quite some time!), perf does not show any performance issues with attribute for me. It only corresponds to ~1.3% of the CPU cycles. `QTextLine::layout_helper` on the other hand consumes ~86.2% of the CPU cycles! So now again I wonder: why are you seeing such drastically different results - these orders of magnitude I cannot easily explain by differences in valgrind vs. perf. REPOSITORY R39 KTextEditor REVISION DETAIL https://phabricator.kde.org/D11487 To: jtamate, #frameworks, #kate Cc: dhaumann, mwolff, cullmann, michaelh, kevinapavew, ngraham, demsking, sars
