Le 24/07/2016 à 17:45, Scott Kitterman a écrit : > I can reproduce what's in your screen cast. Every single version of PyQSci I > try does exactly that regardless of Python version, Qt (4 or 5) version, or > QScintilla version.
I already reassigned to bug to SciTE, since the same issue appears with SciTE (and SciTE is THE app using the Scintilla lib). > I think your screen cast shows normal behavior. Definitely not. When typing characters, the cursor should visually be right after the character typed, at max a few pixels on the right. In the screencast, we can see that the cursor gets further away from the typed character each time. After a string of a dozen characters, the cursor seems to be very far in void space, while it should be right after the last letter. >From what I observed, certain letters do positive offset (cursor goes further on the right), while some letters do negative offset (cursor goes further on the left)... I can't say exactly when this regression started, but it was probably within the few previous weeks on unstable. I can say that Scintilla and QScintilla don't show the problem on stable. > I don't think your test case demonstrates it's a QScintilla problem. Agreed, the report has been reassigned to SciTE. However, it's worth nothing that using a monospaced font (Ctrl+F11 in SciTE, or adding "sci.setFont(QFont('monospace'))" in the example python script) may give a normal result, i.e. no offset incrementally moving away the cursor from its character. Unexplainably, on my machine, it also depends on the desktop used (problems with XFCE, no bug with fluxbox/openbox).