Archive-Date: Fri, 03 May 1996 23:12:00 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Fri, 3 May 1996 17:01:01 -0400 Message-ID: <96050317010142@aalix.pharmark.com> From: george@aalix.pharmark.com Reply-To: vms-emacs-testers@e.kth.se To: VMS-EMACS-TESTERS@e.kth.se Subject: .emacs Hi, In VMS, do we have .emacs file like UNIX. If it has, what is it name, and which directory should it stay? Is it the same format as UNIX one? Thanks a lot!! George ================================================================================ Archive-Date: Sat, 04 May 1996 22:00:53 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Sat, 04 May 1996 03:27:29 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A1CF2.163CBBA7.7@e.kth.se> Subject: Re: .emacs >From: george@aalix.pharmark.com > In VMS, do we have .emacs file like UNIX. If it has, what is it name, and >which directory should it stay? Is it the same format as UNIX one? Thanks >a lot!! The name is .emacs, it should reside in SYS$LOGIN: (your home directory) and it's should look almost exactly the same as on Unix. Just try it out. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Mon, 20 May 1996 16:46:51 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Mon, 20 May 1996 09:26:11 -0500 (CDT) From: GNU%PARTEE@Inland.com Reply-To: vms-emacs-testers@e.kth.se Subject: I found a problem with sysdep.c (here's the patch) To: vms-emacs-testers@e.kth.se Message-ID: <01I4X9OSMZB68XEJGR@research.INLAND.COM> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT I found a problem with sysdep.c when I was using the following configuration on a VAX vms (v6.2) and the VAX c compiler (v3.2-044) Configuration: --PREFIX=EMACS_DEV:[GNU] --SINGLE-TREE=EMACS_DEV:[GNU] --EXEC_PREFIX=EMACS_DEV:[GNU] --WITH-TCPIP=NO It turns out that hostname is never defined in init_system_name. Here is the patch. I really hate doing it this way because it duplicates code from above, but I didn't know how to get around the alloca line. Emacs works fine this way with no network. Another bug was that I couldn't config with the network if I wasn't running DNS. I guess sysdep expects the hostname to contain a period character. I did not fix this because we will be adding DNS to our networking in another month or so. *** sysdep.c;1 --- sysdep.c ************** *** 2373,2378 #endif /* HAVE_GETHOSTNAME */ #ifdef VMS char *sp, *end; if ((sp = egetenv ("SYS$NODE")) == 0) strcpy (hostname, "vax-vms"); else if ((end = index (sp, ':')) == 0) --- 2373,2382 ----- #endif /* HAVE_GETHOSTNAME */ #ifdef VMS char *sp, *end; + #ifndef HAVE_GETHOSTNAME + int hostname_size = 256; /* hostname must still be defined for VMS */ + char * hostname = (char *) alloca (hostname_size); + #endif /* not HAVE_GETHOSTNAME */ if ((sp = egetenv ("SYS$NODE")) == 0) strcpy (hostname, "vax-vms"); else if ((end = index (sp, ':')) == 0) ___________________________________________________________________________ /\ Jeff Delinck (gnu account) |internet: gnu@partee.inland.com /||\ Inland Steel |voice: 219-399-2445 fax: 219-399-5714 \||/ 3210 Watling St. MS 2-465 | This message is printed on re-cyclable East Chicago, IN 46312 USA | magnetic medium. I speak for myself, not the company. ================================================================================ Archive-Date: Mon, 20 May 1996 20:04:10 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Mon, 20 May 1996 19:51:58 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A2A0E.44BAFE20.85@e.kth.se> Subject: Re: I found a problem with sysdep.c (here's the patch) >From: GNU%PARTEE@Inland.com > >I found a problem with sysdep.c when I was using the following >configuration on a VAX vms (v6.2) and the VAX c compiler (v3.2-044) Hmm, it's time I start thinking about inserting some recommendations in the subscription message on where to look for known bugs and such things. This particular bug is well-known, and is described in http://www.e.kth.se/~levitte/gnu/emacs_bugs.html. My fix does not duplicate the code. BTW, I've been somewhat out of GNU hacking busyness for a month or so. For those wondering what happens, I can say that I had a few personal problems (economics, among others) to solve... But now I'm back and will start working on enhancement of Emacs, as far as my current work permits. I've just a few days ago suddenly experienced the problem with hanging subprocesses. This has been reported long ago, but since it didn't happen to me, I've not been able to solve it in any way, but now, I may at last have a chance to find out what is happening. Meanwhile, I think I'm going to make an update, which will most probably be very small, but at least the bug described above will not be present any more (it has really been present for just too long). Enough ranting already... BTW, I plan to move this list to another system (lp.se) during the summer or autumn, since e.kth.se will stop using VMS at the end of this year. Everyone will get two notifications when this happens: one when you get removed from the list at e.kth.se, and one when you're added to the corresponding list on lp.se. These plans are not completelly definite. It completelly depends on what will happen until then. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Tue, 21 May 1996 18:08:06 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Tue, 21 May 1996 17:45:25 +0200 Message-ID: <96052117452514@a400.sternwarte.uni-erlangen.de> From: ai26@a400.sternwarte.uni-erlangen.de (Michael Lemke, Sternwarte Bamberg, Phone: +49-951-9522216) Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Subject: MMS command too long I am trying to build emacs 19.28 and grabbed EMACS-19_28-19960225.TAR-GZ from Sweden. Seems to be ok but if I install it MMS or rather DCL chokes on too long command lines: set def [.lib-src] MMK /macro=("prefix_dev=TOOLS_ROOT_A:","prefix_dir=EMACS","exec_prefix_dev=TOOL S_ROOT_A:","exec_prefix_dir=EMACS.AXP","bindir=TOOLS_ROOT_A:[EMACS.AXP.BIN]","li bdir=TOOLS_ROOT_A:[EMACS.AXP.LIB]",- "archlibdir=TOOLS_ROOT_A:[EMACS.AXP.LIB.EMACS.19_28.ALPHA-DEC-VMS6_2]","MAKE=MMK ") install %DCL-W-TKNOVF, command element is too long - shorten \/MACRO=("prefix_dev=TOOLS_ROOT_A:","prefix_dir=EMACS","exec_prefix_dev=TOOLS_R OOT_A:","exec_prefix_dir=EMACS.AXP","bindir=TOOLS_ROOT_A:[EMACS.AXP.BIN]","libdi r=TOOLS_ROOT_A:[EMACS.AXP.LIB]","archlibdir=TOOLS_ROOT_A:[EMACS.AXP.LIB.EMACS.19 _28.ALPHA-DEC-V %MMS-F-ABORT, For target INSTALL-ARCH-DEP, CLI returned abort status: %X000382A0 . -CLI-W-TKNOVF, command element is too long - shorten Now I could reconfigure my system and create a shorter logical than TOOLS_ROOT_A but I don't want this just to get stupid emacs to install. Defining something like TA temporarily wasn't too good either as it gets copied (compiled even?) into thousands of files. This is a mess. Any suggestions how to fix this? Thanks, Michael -- Michael Lemke Sternwarte Bamberg, University of Erlangen-Nürnberg, Germany (michael@io.as.utexas.edu or ai26@a400.sternwarte.uni-erlangen.de) ================================================================================ Archive-Date: Wed, 22 May 1996 17:16:17 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Return-Path: Message-ID: <9605221806.AA0602@nptpcdns.npt.nuwc.navy.mil> Date: Wed, 22 May 1996 11:35:00 -0400 From: Cobb_David_A Reply-To: vms-emacs-testers@e.kth.se Subject: RE: MMS command too long To: vms-emacs-testers Subject: MMS command too long Date: Tuesday, May 21, 1996 12:25PM I am trying to build emacs 19.28 and grabbed EMACS-19_28-19960225.TAR-GZ from Sweden. Seems to be ok but if I install it MMS or rather DCL chokes on too long command lines: [CLIP] I also had a problem caused by MMK lines too long. In my case I had lengthened them by requesting /LOG/OUTPUT while trying to solve other problems. If I may suggest, could not the desired MACROS be put into a file -- I know MMK accepts the syntax /MACRO=FileSpec. Then they could be as long as anyone desired. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ In real life (which this certainly IS NOT!): David A. Cobb, Code 2253, Naval Undersea Warfare Center Division 1176 Howell Street, Bldg 1171/2, Newport, Rhode Island 02841 Phone: 401.841.4581, Fax: 401.841.2130 ======================================================================== ================================================================================ Archive-Date: Wed, 22 May 1996 17:30:18 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Return-Path: Message-ID: <9605221821.AA0603@nptpcdns.npt.nuwc.navy.mil> Date: Wed, 22 May 1996 11:52:00 -0400 From: Cobb_David_A Reply-To: vms-emacs-testers@e.kth.se Subject: GNU C++ for VMS To: "'LIST: vms-emacs-testers'" I have seen reference on the list to a GNU C++ compiler for VMS. Would someone please point me to a download site? Thanks. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ In real life (which this certainly IS NOT!): David A. Cobb, Code 2253, Naval Undersea Warfare Center Division 1176 Howell Street, Bldg 1171/2, Newport, Rhode Island 02841 Phone: 401.841.4581, Fax: 401.841.2130 ======================================================================== ================================================================================ Archive-Date: Thu, 23 May 1996 00:49:42 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Thu, 23 May 1996 00:42:53 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A2BC9.3DD4B62D.2@e.kth.se> Subject: RE: MMS command too long >From: Cobb_David_A >If I may suggest, could not the desired MACROS be put into a file -- I know >MMK accepts >the syntax /MACRO=FileSpec. Then they could be as long as anyone desired. Yep, I've thought about it. It's just the problem I had not long ago that syopped me from implementing that correctly. Thanks for reminding me. FYI, a change like the one above will require that you start are relatevily new version of MMK (not older than half a year if I recall correctly). Before that, MMK was imcompatable with MMS in this matter. I will do the change during the weekend, where I will also produce a new update. Let's hope I've covered mostly everything by then. I will NOT have the time to check this hanging subprocess problem during the weekend. That will instead be done during next week, when I expect things to be less chaotic than right now. I'm currently playing with MBONE technology, which is completelly new to me, and quite funny. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Thu, 23 May 1996 12:03:20 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Thu, 23 May 1996 11:52:03 +0200 Message-ID: <96052311520297@a400.sternwarte.uni-erlangen.de> From: ai26@a400.sternwarte.uni-erlangen.de (Michael Lemke, Sternwarte Bamberg, Phone: +49-951-9522216) Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Subject: Re: MMS command too long In a message of Thu, 23 May 1996 00:42:53 +0200, received on 23 May 1996, 01:00 Richard Levitte - GNU on VMS hacker wrote to: A mailing list for people testing GNU Emacs on VMS. >>From: Cobb_David_A > >>If I may suggest, could not the desired MACROS be put into a file -- I know >>MMK accepts >>the syntax /MACRO=FileSpec. Then they could be as long as anyone desired. > >Yep, I've thought about it. It's just the problem I had not long ago >that syopped me from implementing that correctly. Thanks for >reminding me. > >FYI, a change like the one above will require that you start are >relatevily new version of MMK (not older than half a year if I recall >correctly). Before that, MMK was imcompatable with MMS in this >matter. FWIW the MMK we had before I put emacs on was not capable of making emacs either. It didn't take a list of targets from the command line. So it was definately a terribly old MMK (2.x). The best solution of course would be for DEC to get rid of the silly 255 char command line limit. Now that they are working on pipes maybe there is a chance... -- Michael Lemke Sternwarte Bamberg, University of Erlangen-Nürnberg, Germany (michael@io.as.utexas.edu or ai26@a400.sternwarte.uni-erlangen.de) ================================================================================ Archive-Date: Wed, 29 May 1996 16:39:50 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 1996 16:28:46 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3104.5FAE2E4B.14@e.kth.se> Subject: YES! Hanging subprocess problem solved! The whole thing started when I decided to store the lower 24 bits of the pid in Emacs' subprocess structure instead of the lower 16 bits. What happened was the if the 24 lowest bits of the subprocess' pid was %HEX 800000 or higher, the subprocess was removed from Emacs' subprocess structures. "Why?" would you ask. Well, very simple. Facts: 1. Those 24 bits are stored in an ELisp integer, which is a 24 bit signed integer. 2. In Unix, a pid is 16 bits. 3. In Emacs, a negative pid was considered to signal a failed subprocess creation. Since the numbers %HEX 800000 and higher would sign-expand to a negative number, Emacs would thing the subprocess creation failed, and would remove the info it had about it, and close the descriptor through which it communicates with the subprocess. The subprocess itself apparently still hangs around, and the lisp code trying to talk to the subprocess will just hang, until you press ^G. The solution to the problem was very simple. Instead of checking if the pid is negative, compare it to -1. Case solved and closed. The patch for this is present in http://www.e.kth.se/~levitte/gnu/emacs_bugs.html, at the end. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Wed, 29 May 1996 16:57:09 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 1996 11:51:56 -0300 (ADT) From: Ben Armstrong Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Subject: Setting up ediff Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII I am an emacs newbie as well as a VMS emacs newbie. I would like to set up ediff so that it works properly on VMS emacs 19.28. However, it attempts to call VMS diff and fails. Fine, I thought. I will just define diff to be a foreign command for a Unix diff I've had kicking around for a while, which appears to be a port of a BSD diff. However, so far I have had no luck. Before defining diff as a foreign command in my login, I get: Errors found in ediff-diff output ... %DCL-W-IVKEYW, unrecognized keyword - check validity and spelling \-C\ After defining in my LOGIN.COM, DIFF=="$somepath:[somedir]diff.exe" I get: Errors found in ediff-diff output ... %DCL-W-IVKEYW, unrecognized keyword - check validity and spelling (that is, I no longer see "\-C\" as the second line of the error) The version number of this diff port is: Revision 2.0 86/09/17 15:40:14 lwall I am wondering if anyone else has succeeded in getting ediff to work and could give me instructions. I can supply further details if needed. Thanks, Ben. -- Ben Armstrong -. Medianet Development Group, BArmstrong@dymaxion.ns.ca `-. Dymaxion Research Limited `- Halifax, Nova Scotia, Canada ================================================================================ Archive-Date: Wed, 29 May 1996 18:33:14 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 96 09:24:31 PDT From: pf%pinegv@Riem.Com Reply-To: vms-emacs-testers@e.kth.se Message-ID: <960529092431.2424@Riem.Com> Subject: hanging subprocess problem To: vms-emacs-testers@e.kth.se Hello, Tried out Richard's patch just now (to process.c). Changed process.c, re-mms'ed. Saw process.c compile. OK that's good. Made a new image; ran that. Never returned from any of the various direds I started. Hmm. Maybe somehow I didn't get the right code. Make a debug version and run that. Set a break at start_process_unwind(). Somewhere at startup it breaks there; must be some kind of subprocess thing. Well the new code is there. Continue. Then fire up a dired. Hangs. And the code _never_ breaks at start_process_unwind. Confused ... pat ================================================================================ Archive-Date: Wed, 29 May 1996 18:42:08 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 1996 18:33:33 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3115.CE3C36E7.3@e.kth.se> Subject: Re: hanging subprocess problem > Maybe somehow I didn't get the right code. Make a debug >version and run that. Set a break at start_process_unwind(). >Somewhere at startup it breaks there; must be some kind of Interesting... >subprocess thing. Well the new code is there. Continue. Then >fire up a dired. Hangs. And the code _never_ breaks at >start_process_unwind. Confused ... *sigh* *grmbllllmmmm* OK, I just tested as well. The only way to "unhang" this was to go to the terminal window and STOP the process doing the directory listing. Back to the drawing table... (BTW, thanks for the quick report) -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Wed, 29 May 1996 18:53:18 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 1996 18:41:56 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3116.FA153AED.67@e.kth.se> Subject: Re: hanging subprocess problem >From: pf%pinegv@Riem.Com >fire up a dired. Hangs. And the code _never_ breaks at >start_process_unwind. Confused ... Did you test on Alpha or on VAX? I've tested on both, and it only hangs on the Alpha instance. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Wed, 29 May 1996 20:12:18 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 1996 20:09:09 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3123.29206A9E.27@e.kth.se> Subject: Re: hanging subprocess problem >From: pf%pinegv@Riem.Com >Made a new image; ran that. Never returned from any of the >various direds I started. Hmm. I found a quirck. The routine in vmsproc.c to read a stream (vms_read_fd) is non-blocking, which means it returns -1 with errno=EWOULDBLOCK on some occations. the function call-process (in callproc.c) wasn't prepared for this, and simply thought something was wrong, and thus stopped the loop, and sat waiting for the subprocess to die. The solution was to check for EWOULDBLOCK. The patch that does this is in http://www.e.kth.se/~levitte/gnu/emacs_bugs.html, at the end Please try it out, Pat (and anyone else), and tell me how it worked out. Thanks. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Wed, 29 May 1996 23:14:15 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 96 14:09:14 PDT From: pf%pinegv@Riem.Com Reply-To: vms-emacs-testers@e.kth.se Message-ID: <960529140914.2502@Riem.Com> Subject: tried Richard's second patch To: vms-emacs-testers@e.kth.se The one to Fcall_process(). Dired still hung. Turned on some of the printf's where you read from the pipe. nread == -1 and errno == 35 (EWOULDBLOCK) as Richard had said. Looked again at the logic of the loop (the while statement) and changed it as follows: Before: while ((nread = tmp_read (fd[0], buf, sizeof buf)) >= 0 && errno != EWOULDBLOCK) After: while ((nread = tmp_read (fd[0], buf, sizeof buf)) >= 0 || errno == EWOULDBLOCK) Didn't make sense to me that you would AND errno with nread >= 0, because I believe the point is that you wanted to continue if this was _not_ the case and errno == EWOULDBLOCK. (Oh, and I of course kept Richard's other changes inside the loop the relevant one of which was to check nread < 0 and if so continue). So this worked. And dired runs _fast_ (as opposed to how things used to be a yr or 2 ago). Remaining problem though. Things are not lined up in the dired buffer. It looks as if some carriage returns need to be inserted because two entries will end up on the same line and then stuff below it will be pushed around to odd postions. And then mysteriously the rebuffer will regain its balance, that it, things will be lined up properly again. pat ================================================================================ Archive-Date: Thu, 30 May 1996 00:19:56 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Thu, 30 May 1996 00:10:24 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3144.DCB26D79.24@e.kth.se> Subject: Re: tried Richard's second patch >From: pf%pinegv@Riem.Com > Before: > while ((nread = tmp_read (fd[0], buf, sizeof buf)) >= 0 > && errno != EWOULDBLOCK) > > After: > while ((nread = tmp_read (fd[0], buf, sizeof buf)) >= 0 > || errno == EWOULDBLOCK) Quite correct. Thanks for the change! This will be included in tomorrows update. > So this worked. And dired runs _fast_ (as opposed to how >things used to be a yr or 2 ago). Yes! At last! > Remaining problem though. Things are not lined up in the dired >buffer. It looks as if some carriage returns need to be inserted >because two entries will end up on the same line and then stuff >below it will be pushed around to odd postions. And then >mysteriously the rebuffer will regain its balance, that it, >things will be lined up properly again. Could you send me an example? That phenomenon sounds very strange to me. There *could* be a problem in vms_read_fd (in vmsproc.c), care to take a look? And now, something completelly different: Since e.kth.se is slowly but steadily dismanteling its VMS cluster, I've started to rework the WWW pages and the structure of the FTP area (or at least, I've started the planning of it). I've started with writing up my plans and wishes in http://www.e.kth.se/~levitte/gnu/gnu.html. I've also moved the current set of pages to http://www.e.kth.se/~levitte/old_gnu/. This means that emacs_bugs.html should now be accessed as http://www.e.kth.se/~levitte/old_gnu/emacs_bugs.html Please read http://www.e.kth.se/~levitte/gnu/gnu.html more than once. It will change while I do all the rework, and will always tell you how it will works out. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Thu, 30 May 1996 03:58:15 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Wed, 29 May 96 18:53:51 PDT From: pf%pinegv@Riem.Com Reply-To: vms-emacs-testers@e.kth.se Message-ID: <960529185351.257b@Riem.Com> Subject: dired buffer contents not lined up properly To: vms-emacs-testers@e.kth.se vms_read_fd() was not executing the translate clause in the 'reading from mbxs' section because read (vms_read_fd) in Fcall_process had no 4th (translate) parameter. I guess in my case there must have been a zero in the next position on the stack. Anyway, add a '1' as a fourth param to the read in the while statment in Fcall_process() and everything gets lined up nicely in the dired buffer. One other thing Richard. Maybe this is just an omission (setting an elisp variable) on my part, but in the dired buffer, as you slide the cursor over the different file names, they don't light up (in green in Emacs on other platforms). MB2 to select and find-file of course does work, but there isn't the visual feedback that the green highlighting normally provides. Is there some elisp variable I need to set? thanx. pat ================================================================================ Archive-Date: Thu, 30 May 1996 10:41:08 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Thu, 30 May 1996 10:38:37 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A319C.9F83BF75.13@e.kth.se> Subject: Re: dired buffer contents not lined up properly >From: pf%pinegv@Riem.Com > vms_read_fd() was not executing the translate clause in the >'reading from mbxs' section because read (vms_read_fd) in >Fcall_process had no 4th (translate) parameter. I guess in my True. Thanks again, Pat, I'm inserting it in the code right now. > One other thing Richard. Maybe this is just an omission >(setting an elisp variable) on my part, but in the dired buffer, >as you slide the cursor over the different file names, they don't >light up (in green in Emacs on other platforms). MB2 to select ... I will check it out. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference. ================================================================================ Archive-Date: Thu, 30 May 1996 19:52:33 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 30 May 1996 19:52:56 +0200 To: vms-emacs-testers@e.kth.se From: gripenstamol@decus.se (Odd Gripenstam) Reply-To: vms-emacs-testers@e.kth.se Subject: Hi I'm trying to use the OO-browser package with VMS Emacs. It works with small test-cases, but I'm having trouble with our real large system. The problem is that string-match aborts with the error message "Stack overflow in regexp matcher". The same thing happens on Linux Emacs 19.27, but not on Linux Emacs 19.30 and DOS Emacs 19.24! The following lisp fragment illustrates the problem. If you remove one of the 'x' it will work fine. (setq string "d::d ( xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx, xxxxxxxxxxxxxxxxxxx, ) {") (string-match "\\`\\s *\\(\\(.\\|\n\\)*\\S \\)\\s *\\'" string) Does anyone know of a way to solve this? I have (naively) tried to compile regex.* from 19.30 with the VMS-kit, but the interface between the modules have changed, so I don't know how much of 19.30 I would have to bring in to VMS to make things compile. /Odd -- Odd Gripenstam Home: +46.8.81.95.58 gripenstamol@decus.se Work: +46.8.700.07.40 odd.gripenstam@om.se Finger gripenstamol@decus.se for my public PGP key. ================================================================================ Archive-Date: Thu, 30 May 1996 19:56:52 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Thu, 30 May 96 10:45:44 PDT From: pf%pinegv@Riem.Com Reply-To: vms-emacs-testers@e.kth.se Message-ID: <960530104544.2735@Riem.Com> Subject: compile-goto-error-error and next-error To: vms-emacs-testers@e.kth.se Compile-goto-error (C-x C-x) and next-error (C-x ~): I seem to be able to get neither of these to work. For example, in a grep output buffer. Several of the items look as follows: > $ grep -n -i ewouldblock *.c /dev/null > callproc.c;8:629: || errno == EWOULDBLOCK) > callproc.c;8:643: continue; /* EWOULDBLOCK */ > process.c;2:2261:#ifdef EWOULDBLOCK > process.c;2:2262: else if (nread == -1 && errno == > EWOULDBLOCK) > process.c;2:2280:#endif /* EWOULDBLOCK */ > process.c;2:2581:#ifdef EWOULDBLOCK I put point on one of these lines. C-x C-x. It should then find-file on this file and go to the correct line number. Instead in the minibuffer I get: Wrong type argument: listp, 677 What have others found? pat ================================================================================ Archive-Date: Fri, 31 May 1996 14:16:01 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Fri, 31 May 96 14:02:34 +0200 Message-ID: <009A328247BDCF40.19C19D1D@om.se> From: "Odd Gripenstam, OM Tech. (+46 8 700 07 40)" Reply-To: vms-emacs-testers@e.kth.se Subject: Re: compile-goto-error-error and next-error To: vms-emacs-testers@e.kth.se grep and compile works for me when I put this in my .EMACS: (setq compilation-error-regexp-alist '(( "^%\\([A-Z0-9]+\\)-[FEW]-\\([A-Z]+\\), \\(.*\\)$" 0 'x compilation-parse-vms-error))) (setq grep-regexp-alist '(("^\\([^( \t\n]+\\)[:( \t]+\\([0-9]+\\)[:) \t]" 1 2 2))) This is with the September update of VMS Emacs. /Odd ================================================================================ Archive-Date: Fri, 31 May 1996 14:22:28 MET-1MET DST Sender: owner-vms-emacs-testers-archiver@ctrl-c.liu.se Date: Fri, 31 May 1996 14:18:14 +0200 From: Richard Levitte - GNU on VMS hacker Reply-To: vms-emacs-testers@e.kth.se To: vms-emacs-testers@e.kth.se Message-ID: <009A3284.77E73600.7@e.kth.se> Subject: Re: compile-goto-error-error and next-error >From: "Odd Gripenstam, OM Tech. (+46 8 700 07 40)" >grep and compile works for me when I put this in my .EMACS: > >(setq compilation-error-regexp-alist > '(( "^%\\([A-Z0-9]+\\)-[FEW]-\\([A-Z]+\\), \\(.*\\)$" 0 'x > compilation-parse-vms-error))) >(setq grep-regexp-alist > '(("^\\([^( \t\n]+\\)[:( \t]+\\([0-9]+\\)[:) \t]" 1 2 2))) > > >This is with the September update of VMS Emacs. As far as I remember, I did a few changes to the "compile" modules during the winter. This means you could win by upgrading. -- +--------------------------------------------------------------------+ ! Richard Levitte, GNU on VMS hacker ! tel: +46-8-26 52 47 ! ! Spannvägen 38, I ! fax: none for the moment ! ! S-161 43 Bromma ! Internet: levitte@e.kth.se ! ! SWEDEN ! ! +-GNUish VMS-+ You may not add me to a commercial mailing list or send me commercial advertising without my consent! See http://www.e.kth.se/~levitte/anti.html for further reference.