Istvan:
I've just compiled the new testing version under Fedora Core 6.
I opened the SciTEGlobal.properties and when I went down to line 253 SciTE
crashed.
OK, the ligature modification couldn't handle Latin1 characters
that take two bytes in UTF-8. Fixed now in CVS and downloads.
I've managed to find out the problem.
THe problem was with the code.page setting.
Maybe a good idea to add some conditions for code page settings depends on OS:
if PLAT_GTK
code.page=65001
LC_CTYPE=en_US.UTF-8
output.code.page=65001
if PLAT_WIN
code.page=0
The problem was a new bug in measuring Latin1 text not in it that
text being Latin1. Setting code.page to UTF-8 (65001) when the text
isn't valid UTF-8 is actually more dangerous than treating it as
Latin1. Try running SciTE from a shell and scrolling over that text:
you should see warnings in the shell complaining about invalid UTF-8
strings.
It could even be seen as a bug that SciTE allows you to run in
UTF-8 mode with text that isn't really UTF-8 but sometimes you need to
edit corrupted files.
Neil
_______________________________________________
Scite-interest mailing list
Scite-interest@lyra.org
http://mailman.lyra.org/mailman/listinfo/scite-interest
