Date: Mon, 30 Mar 92 09:00 EST To: 99clifto AT lab DOT cc DOT wmich DOT edu Cc: djgpp AT sun DOT soe DOT clarkson DOT edu, ericb AT lsid DOT hp DOT com From: Jose DOT Hernandez AT adi_email DOT analog DOT com Subject: Returned mail: Cannot send messa Status: O -------------( Forwarded letter follows )----------------------- Date: Saturday, 28 March 1992 2:37am Date: Sat, 28 Mar 92 02:41:26 EST From: Mailer-Daemon AT nwd2sun1 DOT analog DOT com (Mail Delivery Subsystem) Subject: Returned mail: Cannot send message for 3 hours To: Mailer-Daemon ----- Transcript of session follows ----- ----- Unsent message follows ----- Return-Path: Received: by nwd2sun1.analog.com (4.1/SMI-4.1) id AG10286; Fri, 27 Mar 92 22:52:21 EST Date: Fri, 27 Mar 92 22:52:21 EST From: Mailer-Daemon (Mail Delivery Subsystem) Subject: Returned mail: Cannot send message for 3 hours Message-Id: <9203280352 DOT AG10286 AT nwd2sun1 DOT analog DOT com> To: ----- Transcript of session follows ----- 421 spd.analog.com: Host oceanus.spd.analog.com is down ----- Unsent message follows ----- Received: from adi.analog.com by nwd2sun1.analog.com (4.1/SMI-4.1) id AA09819; Fri, 27 Mar 92 18:57:15 EST Received: from barnacle.erc.clarkson.edu by cs.umb.edu (5.65c/1.34) id AA07643; Fri, 27 Mar 1992 18:50:01 -0500 Received: from sun.soe.clarkson.edu by barnacle.erc.clarkson.edu with SMTP (PP) id <12968-0 AT barnacle DOT erc DOT clarkson DOT edu>; Fri, 27 Mar 1992 18:28:50 -0500 Received: by sun.soe.clarkson.edu (4.1/SMI-4.1) id AA16573; Fri, 27 Mar 92 17:51:49 EST Return-Path: Received: from DUCVAX.AUBURN.EDU by sun.soe.clarkson.edu (4.1/SMI-4.1) id AA16568; Fri, 27 Mar 92 17:51:39 EST Received: from DUCVAX.AUBURN.EDU by DUCVAX.AUBURN.EDU (PMDF #12431) id <01GI59NTXKGW000D5J AT DUCVAX DOT AUBURN DOT EDU>; Fri, 27 Mar 1992 16:49 CST Date: Fri, 27 Mar 1992 16:49 CST From: DARREL HANKERSON Subject: Last word (for now) on globbing To: 99clifto AT lab DOT cc DOT wmich DOT edu Cc: ericb AT lsid DOT hp DOT com, djgpp AT sun DOT soe DOT clarkson DOT edu Message-Id: <01GI59NTXKGW000D5J AT DUCVAX DOT AUBURN DOT EDU> X-Vms-To: IN%"99clifto AT lab DOT cc DOT wmich DOT edu" X-Vms-Cc: IN%"ericb AT lsid DOT hp DOT com",IN%"djgpp AT sun DOT soe DOT clarkson DOT edu" Eric Backus points out that any djgpp-compiled program will handle "response files" (those beginning with "@"). I hadn't even considered checking for this (I feel a bit foolish). This may solve the globbing issue for some programs (certainly for texindex--just use your favorite globber, output to a file, then invoke texindex). This approach does not have the length-limit problem of some of the other suggested solutions. There is a dos-globber in the perl release, and the GNUish glob.c can also be compiled as a nice globber. (ms_sh is another possibility). Thanks to Eric Backus, d.clifton [99clifto AT lab DOT cc DOT wmich DOT edu], and all the others. I've enclosed part of E.B.' note, since the ideas may be of interest to others: -- From E. Backus --------------------------------------------------- Yes, I really do mean go32. It's not exactly documented, so it surprised me as well. Anything you run with go32 will automatically and transparently expand 'response files' on the command line. I've tested and used this feature. To insert globbing at link time means a different crt0.o file. This doesn't sound too hard to me - I might just try to implement this. But if it's implemented in go32 instead of crt0.o, you can have two different go32 programs and decide at RUNTIME whether you want globbing or not. To me this sounds even better than doing it at link time. -- End of message ---------------------------------------------------- Remark: MSC does it at link time. Eric's idea about two different go32 programs has the advantage of not including the globbing code in every exe which may need it. d.clifton also points out this problem, but his recommendation, as I understand it, would require changes to the program you wish to compile. I'd like a simple method which requires no changes to source code. The glob.c that I added to texindex required only changing the name of main(). --darrel hankerson hank AT ducvax DOT auburn DOT edu