Robert Roessler:

> What was the underlying problem?  Was it simple, like the
> non-line-specific marker changes not getting redrawn at all, or
> something more subtle?  I assume it was trickier, since I use markers
> a lot in my client, and never saw any [non]redraw issues...

   The underlying problem was the inaccurate handling of the region
being redrawn. This is (I hope) fixed on Windows native with the
addition of hRgnUpdate and the PaintContains method. It may still be a
problem on other platforms which have not yet implemented
PaintContains. On GTK+/X, painting is not restricted to the update
region so the bug didn't occur and I never investigated GTK+/Windows.

   Redrawing just the line containing the changed marker is also a
worthwhile optimisation.

> Yes, I remember the posted pic of the marker margin not being redrawn
> properly... if I can devise a "cleaner" (to me) fix, would you
> consider it?

   A rearrangement of the notification so that the 'all lines' state
is communicated by something else rather than an out-of-bounds line
number may be OK but I would like to preserve the ability to determine
which line is being changed when possible.

> This looks doable (both "notify these users on any commit" and "notify
> ME when this file changes" sorts of things) - but with admin work and
> scripts on the repository machine... which looks like more than you
> would want to do, even if you had all of the necessary access. :(

   Some SourceForge projects have CVS commits mailing lists but to me
commit streams are too detailed so I prefer running client side diffs
to see changes such as
cvs diff -D yesterday
   or
cvs diff -D 2005-12-26

   Yesterday, CVS thought that LexCaml.cxx had disappeared from the
server: possibly you were updating it but there is nothing in the log.

   Neil

_______________________________________________
Scintilla-interest mailing list
[email protected]
http://mailman.lyra.org/mailman/listinfo/scintilla-interest

Reply via email to