Re: Trying the HEAD (or TIP or whatever)
On 10/10/2008 12:18 AM, Stephen J. Turnbull wrote:
David A. Cobb writes:
> How about capturing the message buffer, which I'm assuming would
> contain the lisp backtrace, into a file?
No, the Lisp stack trace is grabbed directly from the interpreter
stack, since few crashes occur after Lisp errors, at least we hope so.
In any case, opening files from crashing programs is somewhat
problematic. It is quite common for the program to crash again while
trying to shut down more or less sanely from an unrecoverable error.
This is just not a very useful thing to pursue in my opinion,
certainly not for me (I don't have the skills needed to think about
the internal state of a crashing program).
I can see some problems opening a file after a crash. Once upon a time,
one could set the *Messages* buffer so it had an output file (might have
been gnu-emacs, it's been a long while). The last time I tried it, I
couldn't find a way to do it.
What about something like the Common Lisp "dribble" functionality, every
interaction is captured into a file.
> Or, of course, simply writing the backtrace to a file.
xemacs 2> file-of-your-choice
Ah. I've had mixed results with redirections from interactive programs.
I'm glad to hear XEmacs doesn't suffer from those problems.
Alternatively, use the 'lbt' command defined in src/.gdbinit from gdb,
and capture it in the same way as the C stack trace.
Thank you for that one (also).
--
David A. Cobb, semi-retired mainframe T-Rex.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta