Re: Elusive assertion failure related to completion

[martin rudalics]

> I've been trying to debug it, but the backtraces are not always
> identical. One sample follows.

Does the following part show up in all backtraces?

> [Switching to thread 1 (thread 3668.0x2f0)]#0  delete_all_subwindows
> (w=0x33c0a00) at window.c:6314
> 6314      w->total_lines = w->buffer;       /* See
> Fset_window_configuration for excuse.  */

If so, does the bug trigger the same way when you replace that

  w->total_lines = w->buffer;       /* See Fset_window_configuration for 
excuse.  */

line by something like

  if (BUFFERP (w->buffer))
    w->total_lines = w->buffer;       /* See Fset_window_configuration for 
excuse.  */

which is just a wild guess, though.

martin