Fabrice Popineau <Fabrice.Popineau(a)supelec.fr> wrote:
I'm using xemacs-21.5b20 from CVS, native win32. I'm using
gnus and
spambayes. I was wondering why from time to time, some messages got
truncated. I have traced the problem up to call-process-internal. I have
got one such message where the truncation appears and saved it in some
foo.txt file.
I used the following function to test spambayes filter against this
message:
(with-temp-buffer
(let
((default-process-coding-system '(binary . binary))
(debug-windows-io 1))
(insert-file "c:/tmp/foo.txt")
(call-process-region (point-min) (point-max)
"c:/local/python24/scripts/dist/sb_filter.exe" t t nil "-f")
(length (buffer-string))))
It is called the same way as gnus is calling it. If I run this code
straight, truncation occurs. If I edebug-defun the call-process-internal
function and run the same code, stepping by hand into
call-process-internal, no truncation occurs.
Anybody has an idea of the problem? It does not occur very often, but it
is a bit scary when you think about it.
Thanks in advance for any input on this issue,
This sounds just like the Unix email truncation bug that was fixed not
long ago. I wonder if the same logic error was made in the
Windows-specific code. Here is the way I produced the error:
http://list-archive.xemacs.org/xemacs-beta/200407/msg00174.html
and here is the patch that fixed the problem
http://list-archive.xemacs.org/xemacs-patches/200502/msg00051.html
Should a similar change be made to nt_send_process?
--
Jerry James
http://www.ittc.ku.edu/~james/