wresize & resizeterm (Was: derwin problems)

[Roger Gammans]

On Sun, May 05, 2002 at 11:48:25AM +0200, Gernot Hillier wrote:
> BTW: The SIGSEGV you experienced with free and malloc are only some 
> indications of memory corruption caused earlier. You get better bt's when 
> linking to libefence. (LD_PRELOAD will do the trick, no recompile is 
> necessary IMHO).

I've been tracking something related here , last week.
With 5.0, I saw clear memoery corruptionm where a malloc
block  overflows (eg written to above it's end).

5.2-20020119a as mentioned in the debian BTS is certainly more
stable, but I'm still see odd problems with wresize and
resizeterm.

wresize failling inside resizeterm - is particulary annoying,
beacuse I  follow it resizing  my windows anyway, but if
any wresize call fails LINES & COLS aren't setup correctly
so it becomes impossible to newwin the new screen size.

I've been playing with a patch do this before in resizeterm
before calling wresize so it is always done. 

I've also been considering something like a new window flag
to indicate it's owner resized.

However, while these changes seem to make it more stable I'm not 
sure they actually fix the error.

> > > I've created a fairly simple program that does this. If you would like I
> > > can attach it (I wasn't sure if you can have attachments in this list).
> >
> > I've seen a couple of related reports in Debian, but don't have a simple
> > program to exercise it.

Im going to try to create a simplified version of my app to 
see if I can trigger it with that which I'll make available.

Did I see someone mention a trace replay tool in the debian bug a
archives though - is that available?

-- 
Roger.
Master of Peng Shui.  (Ancient oriental art of Penguin Arranging)