koloska wrote: > Hello, > > just another curiosity from me -- I hope you don't find the > performance boring ;-) > > Since bidi was not supported by the text widget, to display arab text > I have used RS arblish routines with great success. > > But now I have found, that on some windows XP computers the text > widget seems to be confused by some bidi ... > > I have tried: > 8.5b1 (tclkit) > 8.5b2 (kbskit) > > (have to test with earlier and later versions, but for earlier > version, I have to write a test program, cause the program uses Ttk) > > The text is already reverted to get RL and this works fine on my winXP > and my linux box. But there are two computers that revert the > reversion for some characters. It seems to be related with the arabic > '?'. The string is something like "?dcba" (this is already reverted > and actually in arabic) and will be shown as "cbad?" If I click on the > word (or make a selection) the characters change their places but > retain their sequence. > > So my question: > > - I have read in the changelogs that bidi may be supported for the > text widget, but haven't found any preconditions ("language pack hast > to be installed" or such) May? Where did you read that? > - Is it possible to disable the bidi rendering and just rely on the > normal state of affairs until bidi is implemented fully? Don't hold your breath. It's a difficult problem with the current structure of the text-related widgets and their assumptions. Feel free to volunteer and fix it. I'd also like the hack with end always being \n to go away in the text widget. :) I know from experience that writing a *good* text widget is very difficult. I'm still working on mine for NexTk. George