[PATCH] xemacs-21.5-clean: call-process-internal workaround for infinite hang of gpg and w3m

Hi All! I have noticed for many months that sometimes file encoding or decoding via program "gpg" would hang. Likewise I saw hangs of program "w3m" when reading HTML mail in gnus. Finally I found these to be related and dependent on the size of input data to them. That is to say I found reliable test cases. Now I have been running with following bad workaround (a pause of 50ms before sending EOF) for some months but have not found time to investigate towards a real solution. Can someone please help with this issue? Best regards! Adrian xemacs-21.5-clean source patch: Diff command: cvs -f -z3 -q diff -u -w -N Files affected: lisp/process.el =================================================================== RCS Index: lisp/process.el =================================================================== RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/process.el,v retrieving revision 1.20 diff -u -w -r1.20 process.el --- lisp/process.el 6 Jul 2006 17:07:41 -0000 1.20 +++ lisp/process.el 13 May 2007 21:18:21 -0000 ＠＠ -278,11 +278,18 ＠＠ ;; whole function, and the unwind-protect will ;; kill the process. (Hence the documented semantics ;; of SIGINT/SIGKILL.) - (let (eof-sent) + (let (eof-sent + (proc-mark 0)) (condition-case nil (progn (when inbuf (process-send-region proc start end inbuf)) + ;; (message "process-buffer:%S" (process-buffer proc)) + ;; (message "process-buffer:%S" (process-buffer proc)) + ;; (message "(point-max buffer) before:%S" (point-max buffer)) + ;; (while (> (process-mark proc) proc-mark) + (sit-for 0.05) + ;; (message "proc-mark:%S:process-mark after:%S:" proc-mark (setq proc-mark (process-mark proc)))) (process-send-eof proc) (setq eof-sent t) (while t -- Adrian Aichner mailto:adrian＠xemacs.org http://www.xemacs.org/ _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta