bug#14630: 24.3.50; emacs_backtrace.txt

[martin rudalics]

> The backtraces reported by Drew consistently point to this line in
> w32fns.c:
>
>        form.rcArea.top += WINDOW_HEADER_LINE_HEIGHT (w);
>
> i.e. to whatever happens in the expansion of
> WINDOW_HEADER_LINE_HEIGHT.

But quite a lot can happen in this expansion.  Can this fail in
CURRENT_HEADER_LINE_HEIGHT?  Drew - do you use header lines in the first
place?

> The XBUFFER part there was already handled
> by the BUFFERP condition, so the only one remaining is XWINDOW.  Which
> is why I added WINDOWP.

You mean FRAMEP?

>> Maybe we should start replacing BUFFERP by BUFFER_LIVE_P.  BUFFERP (like
>> WINDOWP and FRAMEP) is IMHO harmful virtually everywhere.
>
> I didn't touch any BUFFERP or related macro in the last change.

I know.  I meant that instead of BUFFERP (w->contents) we could check
BUFFER_LIVE_P (XBUFFER (w->contents)) there.

martin