Re: xemacs crash
On Fri, Oct 8, 2010 at 10:54 PM, Stephen J. Turnbull <stephen(a)xemacs.org> wrote:
Samuel Bronson writes:
> Fatal error.
Huh, no error code or line number, etc?
> # bind (print-message-label)
> error-message-string((wrong-type-argument
>
> and it breaks off there, as if it had either crashed in the midst of
> printing this line or not been able to flush the rest.
That does look look a crash in the error-reporting mechanism (perhaps
during an attempted shutdown after an earlier crash).
Well, what I mean is that I got a "doctor watson" dialog too, so maybe
xemacs crashed again while displaying the truncated stack trace shown
here. (Or maybe it's that broken IO buffering behavior I've heard
MSVCRT has, and emacs just punted after it had printed the whole
thing, but some of it was still stuck in the buffer when the outermost
SEH handler got invoked?)
Is this repeatable?
I'm not sure.
> VC++ could only give me the following "stack trace";
I'm not sure how
> reliable it is. (I do also have a minidump if someone wants to take a
> look at that, but it's 10 megs, so I'm not attaching it to this
> message.)
> > xemacs.exe!011a7fbe()
> xemacs.exe!0118ebc1()
This is strange. Windows backtraces usually have stack frame
information, I thought, such as names of called functions and
arguments. Without that, I don't think we can do much.
Are you stripping symbols out of the binary?
Stripping the binary? I haven't even figured out how to build my own
yet! This is the one from the installer in the xemacs ftp tree.
Anyway, as the configuration information I pasted shows, this was
built with "cl", MS's C compiler (or a drop-in replacement?). MS's
compiler hasn't supported having debugging symbols in the executable
file itself for quite some time. Instead, they are kept in separate
files, with the extension ".pdb" (assuming they were built; I don't
think this is mandatory *yet*). Anyway, judging by the last few bytes
of the executable, they were in this case:
00b80000: 4e42 3130 0000 0000 d345 394a 0100 0000 NB10.....E9J....
00b80010: 433a 5c58 456d 6163 7342 7569 6c64 5c78 C:\XEmacsBuild\x
00b80020: 656d 6163 732d 3231 2e35 2e32 395c 7372 emacs-21.5.29\sr
00b80030: 635c 7865 6d61 6373 2e70 6462 00 c\xemacs.pdb.
(I'm slightly surprised that they're in a file called xemacs.pdb and
not temacs.pdb.)
Anyway, suffice to say that this file was not installed along with the
executable. And yes, MS's tools would have been able to find it there
if it had -- they don't assume that the symbols are still at the path
they were built at. (Unlike, say, `find-function' w.r.t. elisp
files...)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta