Thu 26 Nov 2015 03:35:55 PM UTC, original submission:
What I mean is: when you hit Enter somewhere in a group of lines that are matched by a multiline regex, and this causes part of the group to (correctly) become uncoloured, then adding back characters that should cause the whole group to become coloured again, do /not/ have this effect.
To reproduce, compile nano with just --disable-libmagic. Don't use --enable-debug, because then the bug will be triggered always. Then run:
MALLOC_PERTURB_=0 src/nano +2,22 ChangeLog
Type: <Enter> <Enter> <Up> xxx.
After the second Enter, the "(do_redo)" correctly becomes uncoloured. But the xxx then do not restore the colour. However, when running:
MALLOC_PERTURB_=242 src/nano +2,22 ChangeLog
and typing the same, the colour /is/ restored.
I have a patch that fixes bug #46543, but it aggravates this bug, triggering it for all values of MALLOC_PERTURB_. That's not good. (So maybe the patch isn't a "fix", just an avoidance strategy -- the actual uninitialization isn't fixed.)
|