From jdennis@remus.ultranet.com Wed Sep 18 15:14:48 1996 X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil] [nil "Wed" "18" "September" "1996" "17:48:37" "-0400" "John R. Dennis" "jdennis@ultranet.com" nil "27" "Call Process under NTemacs" "^From:" nil nil "9" nil nil nil nil] nil) Received: from perseus.ultra.net (perseus.ultra.net [199.232.56.6]) by june.cs.washington.edu (8.7.5/7.2ju) with ESMTP id PAA19578 for ; Wed, 18 Sep 1996 15:14:47 -0700 Received: from dakota.ultranet.com (dakota.ultranet.com [205.246.13.36]) by perseus.ultra.net (8.7.4/dae0.6) with SMTP id SAA15216 for ; Wed, 18 Sep 1996 18:14:44 -0400 (EDT) Received: by dakota.ultranet.com with Microsoft Mail id <01BBA58D.1C673C00@dakota.ultranet.com>; Wed, 18 Sep 1996 18:13:36 -0400 Message-ID: <01BBA58D.1C673C00@dakota.ultranet.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit From: "John R. Dennis" To: "'voelker@cs.washington.edu'" Subject: Call Process under NTemacs Date: Wed, 18 Sep 1996 17:48:37 -0400 Geoff: I'm running emacs 19.31 on NT 4.0. When a lisp program invokes call-process sending a region as stdin to the process, it fails with an invalid argument error. I stepped though the lisp debugger and saw something very interesting. Just before call-process is invoked the arguments look just as I expect them to. Then I step once in the debugger, the call-process generates an "Invalid Argument" error, and (this is the interesting part) the arguments to call-process in the debugger have now changed from what they were before the call to call-process was made. The argument list now includes "C:/TEMP/emf00092". I'm going to guess that this temp file (which exists) was created to hold the regions data, that instead of passing the region as stdin, its now passed as a file argument. I assume this was done by the C code as a workaround and hence is why the argument list is different before and after the call to call-process. I also assume the reason I got an "invalid argument" was because the temp file name uses the Unix / as a path separator instead of backslash. I searched the source code for something that would be creating a temp file with the prefix "emf" but I didn't find anything. Does this ring a bell with you? Do you have a workaround? Do you have any comments? Thanks, John Dennis jdennis@ultranet.com From owner-ntemacs-users@june Mon Aug 12 21:06:58 1996 X-VM-v5-Data: ([nil nil nil nil t nil nil nil nil] [nil "" "12" "August" "1996" "23:24:37" "-0400" "Michael Welsh Duggan" "md5i@schenley.com" "" "38" "start-process woes" "^From:" nil nil "8" nil nil nil nil] nil) Received: from joker.cs.washington.edu (joker.cs.washington.edu [128.95.1.42]) by june.cs.washington.edu (8.7.5/7.2ju) with SMTP id VAA28702 for ; Mon, 12 Aug 1996 21:06:58 -0700 Received: from june.cs.washington.edu (june.cs.washington.edu [128.95.1.4]) by joker.cs.washington.edu (8.6.12/7.2ws+) with ESMTP id VAA22963 for ; Mon, 12 Aug 1996 21:06:52 -0700 Received: from maru.schenley.com (maru.schenley.com [206.62.188.70]) by june.cs.washington.edu (8.7.5/7.2ju) with SMTP id UAA25963 for ; Mon, 12 Aug 1996 20:25:21 -0700 Received: by maru.schenley.com from localhost (router,SLmailNT V2.0); Mon, 12 Aug 1996 23:24:39 EDT Received: by maru.schenley.com from MARU (127.0.0.1::mail daemon; unverified,SLmailNT V2.0); Mon, 12 Aug 1996 23:24:39 EDT Message-ID: Lines: 38 X-Mailer: Gnus v5.2.37/Emacs 19.32 From: "Michael Welsh Duggan" Sender: "Michael Welsh Duggan" To: ntemacs-users@cs.washington.edu Subject: start-process woes Date: 12 Aug 1996 23:24:37 -0400 I have had some problems with start-process for quite a while. In order to test this, I ran the following piece of code from a lisp-interactive buffer which which thought that the emacs-19.32/src directory was its home (this was culled from the actual call grep.el makes): (BTW, the variable `grep' held the result of (get-buffer "*grep*"), just to be pedantic (this is what grep.el places in there).) (start-process "grep" grep "C:\\WINNTSRV\\system32\\cmd.exe" "/c" "grep -n start-process *.c NUL") This comes back in the *grep* buffer with the quite reasonable response of: ---------------------------------------------------------------------- fileio.c:416:The `call-process' and `start-process' functions use this function to\n\ process.c:1004:DEFUN ("start-process", Fstart_process, Sstart_process, 3, MANY, 0, process.c:3770:The value takes effect when `start-process' is called."); Process grep finished ---------------------------------------------------------------------- However, when I execute this a second time (or any subsequent times), I get the following: ---------------------------------------------------------------------- Bad command or file name Process grep finished ---------------------------------------------------------------------- What's up with this? It makes it difficult to do any two M-x grep or M-x diffs in a single session. -- Michael Duggan (md5i@schenley.com)