bug#15213: 24.3.50; emacs_backtrace.txt

[Drew Adams]

Thanks for fixing the bug(s).

15213 and 15183 were merged.  Not sure whether your "It's not related."
means that those are two different bugs or that your previous fix for
15183 is unrelated to 15213.

Anyway, I just wanted to point out, relative to your guess that this
might be a result of my using a file compiled in an earlier Emacs
release, that 15183 was from emacs -Q, and IIRC none of my code was
ever loaded in that session.  The crash came very soon after emacs -Q.

(Also, FWIW, I have no code that calls `temp-output-buffer-show'.
But of course it might be called by other code, including vanilla code
from a previous release, as you suggest.)

HTH.

>  > This is the same as #15183, we hope Martin fixed that in revision
>  > 14005 (the backtrace was produced from revision 113986).
> 
> It's not related.  Basically, what happens here is that the window
> produced by `display-buffer' for `temp-output-buffer-show' is not live.
> This is a case `temp-output-buffer-show' doesn't handle, at least not on
> trunk:
>        window = display_buffer (buf, Qnil, Qnil);
> 
>        if (!EQ (XWINDOW (window)->frame, selected_frame))
>       Fmake_frame_visible (WINDOW_FRAME (XWINDOW (window)));
> 
> But `temp-output-buffer-show' is obsolete since 24.3.  I suppose Drew
> runs code byte-compiled with some pre 24.3 Emcas on trunk and either
> does not produce a new window in `display-buffer' or delete it before
> `temp-output-buffer-show' can deal with it.