path: root/vms/vms-notes

Thu Nov 22 18:18:47 IST 2012
============================

Per Anders Wallin, the HP VMS porting guide is available at

	http://h71000.www7.hp.com/portability/portingguidelines.html

This file documents this fact, and anything else of interest.
From wb8tyw@qsl.net  Wed Dec 11 20:03:44 2013
Return-Path: <wb8tyw@qsl.net>
Received: from skeeve.com (skeeve.com [127.0.0.1])
	by skeeve.com (8.14.4/8.14.4/Debian-2ubuntu2.1) with ESMTP id rBBI3gNX002464
	for <arnold@localhost>; Wed, 11 Dec 2013 20:03:43 +0200
X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on
	frenzy.freefriends.org
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM
	autolearn=ham version=3.3.1
X-Envelope-From: wb8tyw@qsl.net
X-Envelope-To: <arnold@skeeve.com>
Received: from frenzy.freefriends.org [66.54.153.139]
	by skeeve.com with IMAP (fetchmail-6.3.21)
	for <arnold@localhost> (single-drop); Wed, 11 Dec 2013 20:03:43 +0200 (IST)
Received: from Encompasserve.org (Eisner.encompasserve.org [67.53.90.116])
	by freefriends.org (8.14.6/8.14.6) with ESMTP id rBB5jaW9020988
	for <arnold@skeeve.com>; Tue, 10 Dec 2013 22:45:40 -0700
Received: from [192.168.0.101]
 (173-19-202-206.client.mchsi.com [173.19.202.206])
 by Encompasserve.org (PMDF V6.6 #13031)
 with ESMTPSA id <01P1SYL5LELG002FZE@Encompasserve.org> for arnold@skeeve.com;
 Tue, 10 Dec 2013 23:45:35 -0600 (CST)
Date: Tue, 10 Dec 2013 23:45:36 -0600
From: "John E. Malmberg" <wb8tyw@qsl.net>
Subject: Re: Some questions...
In-reply-to: <201312110426.rBB4QXXc003297@skeeve.com>
To: Aharon Robbins <arnold@skeeve.com>
Cc: wb8tyw@qsl.net
Message-id: <52A7FC00.2040905@qsl.net>
MIME-version: 1.0
Content-type: text/plain; charset=ISO-8859-1; format=flowed
Content-transfer-encoding: 7BIT
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:17.0) Gecko/20130215
 Thunderbird/17.0.3
References: <529A9537.7090404@qsl.net> <201312110426.rBB4QXXc003297@skeeve.com>
Status: RO

On 12/10/2013 10:26 PM, Aharon Robbins wrote:
> Hi. Some questions about your changes.
>
> 1. I notice that there is at least one instance of #if __VMS, where as
> it seems most code uses #if VMS.  Is that on purpose?  If not, which
> is correct?

VMS is a macro historically pre-defined by C compilers on C before the 
ANSI rules.

Under strict ANSI rules, a standards compliant compiler is not allowed 
to pre-define macros that are not prefixed by "__".

Which means that you can not count on the VMS macro being defined unless 
you do it yourself.

So __VMS is the ANSI correct definition.  But it would be a lot of work 
to change all the code.

The vms_config.h that is now generated has these lines in it to make 
sure that VMS is defined.

#ifdef __VMS  /* ANSI compliant */
#ifndef VMS /* PRE-ansi */
#define VMS 1
#endif
#endif

This may be needed for other architectures.  It only shows up as an 
issue if you set a "strict" compliance flag on the compile command.

> 2. For the include of the main wrapper.  Gawk already has a routine
> os_arg_fixup which is intended to manipulate argc and argv as needed
> by an OS-specific routine before main() parses it.  A quick look at
> the wrapper looks to me like it could be made to fit within that framework,
> which would keep #if and the include out of the mainline code.

I can look at that.  I can also look to see if I can move the logic from 
vms_crtl_init.c into it in a way that would sidestep that it has no 
copyright notice at all.  The result would be code that is almost 
identical to code that is in the official Perl distribution.

GNV was started as a user project and when the VMS POSIX product was 
dropped and they needed a Unix like environment, Digital took over the 
GNV product, but did not really do much to keep it up to date.

The routine in vms_crtl_init.c is actually called before main() is 
called, so just linking a module with it in is all that it takes to 
activate the change.  Unfortunately I do not know how to get it to 
magically fix up the argv array.

> If that looks reasonable to you, can you make that change?  You can send
> it as a diff against what you currently have already done.

I will look at it tomorrow night.  I also found another change needed 
for vms_args.c to work properly under a bash shell.

> I'm sorry about the hassle on the copyrights, but I really have to play
> this game by the FSF rules.  I hope we can get it clarified soon
> so that I can include the files in the dist.

I hope so.  I have not seen a reply from Karl yet.  He did say that 
there are other options.

This would clear the way for getting the similar code accepted into the 
Bash and coreutils repositories.  I have a report that the first bash 
4.3 beta built on VMS and looked real good on the test run.

I am currently working on adapting the packaging scripts from bash and 
coreutils to gawk.  Each product has had it unique features, so I have 
not yet come up with a generic set of scripts.

This is a preview what will be coming:

backup_gawk_src.com - Builds a VMS equivalent to tar archive.  By 
placing the source in the PCSI kit, it covers the requirement of 
providing source somewhere with the binary.

build_gawk_pcsi_desc.com - Creates a VMS PCSI package manifest file.

build_gawk_pcsi_text.com - text output by the VMS PCSI package installer.

compare_gawk_source.com - I have the source checked out on an NFS 
server.  The backup program needs me to copy it to a VMS volume first. 
This procedure makes sure that the two copies are the same and ready for 
backup.

gawk_alias_setup.com - Sets up gawk and awk links to gnv$gawk.exe. 
Needed to repair when sins of the past partially break an install.

gawk_verb.cld - for gawk to use gnv$gawk.exe as an image.  Different 
than gawk.cld.  To properly add an image to the VMS command table, it 
needs a prefix.  I have "GNV$" registered as a prefix for this purpose.

gnv_gawk_kit_name.com - Calculates the VMS PCSI package name, used for 
both file names and file contents.

gnv_gawk_startup.com - Run at VMS startup to make sure that the 
environment is set up properly for gawk.

pcsi_gawk_file_list.txt - Source file for build_gawk_pcsi_desc.com.

remove_old_gawk.com - Needed to properly clean up sins of the past. 
Only touches the the GNV provided gawk and awk programs.

stage_gawk_install.com - Does a pre-install into a fake "root" directory 
tree for the kitting procedure to pull files out of.

Regards,
-John

From wb8tyw@qsl.net  Thu Dec 12 12:21:57 2013
Return-Path: <wb8tyw@qsl.net>
Received: from skeeve.com (skeeve.com [127.0.0.1])
	by skeeve.com (8.14.4/8.14.4/Debian-2ubuntu2.1) with ESMTP id rBCAIn2l004322
	for <arnold@localhost>; Thu, 12 Dec 2013 12:21:55 +0200
X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on
	frenzy.freefriends.org
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM
	autolearn=ham version=3.3.1
X-Envelope-From: wb8tyw@qsl.net
X-Envelope-To: <arnold@skeeve.com>
Received: from frenzy.freefriends.org [66.54.153.139]
	by skeeve.com with IMAP (fetchmail-6.3.21)
	for <arnold@localhost> (single-drop); Thu, 12 Dec 2013 12:21:55 +0200 (IST)
Received: from Encompasserve.org (Eisner.encompasserve.org [67.53.90.116])
	by freefriends.org (8.14.6/8.14.6) with ESMTP id rBC6FboY024650
	for <arnold@skeeve.com>; Wed, 11 Dec 2013 23:15:41 -0700
Received: from [192.168.0.101]
 (173-19-202-206.client.mchsi.com [173.19.202.206])
 by Encompasserve.org (PMDF V6.6 #13031)
 with ESMTPSA id <01P1UDWPUU7M002J4J@Encompasserve.org> for arnold@skeeve.com;
 Thu, 12 Dec 2013 00:15:35 -0600 (CST)
Date: Thu, 12 Dec 2013 00:15:38 -0600
From: "John E. Malmberg" <wb8tyw@qsl.net>
Subject: Re: Some questions...
In-reply-to: <201312110755.rBB7tIrb026097@freefriends.org>
To: arnold@skeeve.com
Cc: wb8tyw@qsl.net
Message-id: <52A9548A.5060604@qsl.net>
MIME-version: 1.0
Content-type: multipart/mixed; boundary="Boundary_(ID_0V6xQMBg8dMk2LzZyR0ioQ)"
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:17.0) Gecko/20130215
 Thunderbird/17.0.3
References: <529A9537.7090404@qsl.net> <201312110426.rBB4QXXc003297@skeeve.com>
 <52A7FC00.2040905@qsl.net> <201312110755.rBB7tIrb026097@freefriends.org>
Status: RO
X-Status: A

This is a multi-part message in MIME format.

--Boundary_(ID_0V6xQMBg8dMk2LzZyR0ioQ)
Content-type: text/plain; charset=ISO-8859-1; format=flowed
Content-transfer-encoding: 7BIT

I moved the vms_gawk_main_wrapper.c code into gawkmisc.vms so that file 
is not needed anymore.

I also discovered that we were not building with exact case public 
symbols, so I fixed up the source to build with exact case symbols. 
This will be more important when building dynamic plug-ins, and in the 
future if and when the other related packages can be built with it.

This caused small changes to most of the source files in VMS, as all the 
system services routines were in the wrong case.

I see that just about everything has been backed out of git until the 
legal stuff is resolved.

This is what I currently have for io.c changes based on a branch I just 
made against master.

And this is what the diff for the vms/gawkmisc.vms changes are, which 
eliminates the change to main.c.

Regards,
-John



On 12/11/2013 1:55 AM, arnold@skeeve.com wrote:
> Hi.
>
> Thahks for the answers. I will address some of this later; I'm at work now.
>
> I would like to emphasize that, from my perspective, the priorities are:
>
> 1. Code changes (os_arg_fixup, etc.).
> 2. Documentation revisions.
> 3. The other VMS-only goodies that you described.
>
> The documentation revisions are not optional; I have to have them.
>
> But, as I said, it's enough if you give me plain text. I can do the
> markup.  If the markup is what's holding you back then don't bother,
> just send me text, with some kind of clear indication of the sectioning /
> subsectioning (if any), and I'll do the markup.
>
> Karl replied further to me; I don't know if it was on purpose or
> by accident that he didn't include you. I will include you on the
> further correspondance that I need to send.
>
> Believe me, I very much want to move this forward, but I really do
> have to play by the rules.
>
> Thanks,
>
> Arnold
>


--Boundary_(ID_0V6xQMBg8dMk2LzZyR0ioQ)
Content-type: application/gzip; name=gawk_io_c_vms.patch.gz
Content-transfer-encoding: base64
Content-disposition: attachment; filename=gawk_io_c_vms.patch.gz
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--Boundary_(ID_0V6xQMBg8dMk2LzZyR0ioQ)
Content-type: application/gzip; name=gawk_gawkmisc_vms.gdiff.gz
Content-transfer-encoding: base64
Content-disposition: attachment; filename=gawk_gawkmisc_vms.gdiff.gz
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--Boundary_(ID_0V6xQMBg8dMk2LzZyR0ioQ)--

From wb8tyw@qsl.net  Sun Dec  8 19:58:11 2013
Return-Path: <wb8tyw@qsl.net>
Received: from skeeve.com (skeeve.com [127.0.0.1])
	by skeeve.com (8.14.4/8.14.4/Debian-2ubuntu2.1) with ESMTP id rB8Hw8Y1002346
	for <arnold@localhost>; Sun, 8 Dec 2013 19:58:10 +0200
X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on
	frenzy.freefriends.org
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM
	autolearn=ham version=3.3.1
X-Envelope-From: wb8tyw@qsl.net
X-Envelope-To: <arnold@skeeve.com>
Received: from frenzy.freefriends.org [66.54.153.139]
	by skeeve.com with IMAP (fetchmail-6.3.21)
	for <arnold@localhost> (single-drop); Sun, 08 Dec 2013 19:58:10 +0200 (IST)
Received: from Encompasserve.org (Eisner.encompasserve.org [67.53.90.116])
	by freefriends.org (8.14.6/8.14.6) with ESMTP id rB84BBd2018585
	for <arnold@skeeve.com>; Sat, 7 Dec 2013 21:11:14 -0700
Received: from [192.168.0.101]
 (173-19-202-206.client.mchsi.com [173.19.202.206])
 by Encompasserve.org (PMDF V6.6 #13031)
 with ESMTPSA id <01P1OOF0VUTE001YI1@Encompasserve.org> for arnold@skeeve.com;
 Sat, 07 Dec 2013 22:11:09 -0600 (CST)
Date: Sat, 07 Dec 2013 22:11:05 -0600
From: "John E. Malmberg" <wb8tyw@qsl.net>
Subject: Re: VMS patch for config.h generation, passing tests.
In-reply-to: <52A277C3.303@qsl.net>
To: "John E. Malmberg" <wb8tyw@qsl.net>
Cc: arnold@skeeve.com, r.pat.rankin@gmail.com, anders_s_wallin@yahoo.se
Message-id: <52A3F159.9000607@qsl.net>
MIME-version: 1.0
Content-type: text/plain; charset=ISO-8859-1; format=flowed
Content-transfer-encoding: 7BIT
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:17.0) Gecko/20130215
 Thunderbird/17.0.3
References: <52A17089.9060701@qsl.net>
 <201312060840.rB68eH2F029601@freefriends.org> <52A277C3.303@qsl.net>
Status: R
X-Status: A

I have merged in your changes with my local copy, but not sure yet what 
to do to get the git status happy.

Unfortunately the VMS 8.3 Alpha manyfile test failed.  Closing 2 files 
in io.c was not good enough.  I am trying again with closing 4 files.
Still failed.  I am going to have to add some diagnostics to identify 
the failure.

And the beginfile1 test also is failing on VMS 8.3 Alpha.  I will look 
at that after I resolve the manyfile issue.

I also did not notice that ofmta test failed on VMS 7.3 VAX.  The 
floating point number is rounding differently.  I get the same result 
with the replacement SNPRINTF or the VMS provided SNPRINTF.

TOAD> diff ofmta.ok sys$disk:[]_ofmta.tmp
************
File SRC_ROOT:[GAWK.TEST]OFMTA.OK;1
     5   4.234
     6   4.2345 hi
******
File LCL_ROOT:[GAWK.TEST]_OFMTA.TMP;2
     5   4.235
     6   4.2345 hi
************

I tried that build again with #undef HAVE_VPRINTF, and it failed.

 From eval.c:
  I1        8733 #ifndef HAVE_VPRINTF
  I1        8734 #error "you lose: you need a system with vfprintf"
                 .1
%CC-E-ERRORMESSAGE, (1) #error "you lose: you need a system with vfprintf"

Note that the error message has the wrong routine name.

I have looked over the manual, and will try to get some detailed changes 
to it.

Some preliminary things to help collect my thoughts.

As the MMS program has a lot of issues with ODS-5, Some open source 
programmers have switched to Madgoat Make, known as MMK from 
https://github.com/endlesssoftware/mmk .  MMK uses most unmodified MMS 
description files and handles ODS-5 volumes better than MMS does.

The optional POSIX subsystem is no longer supported on VMS.  While I 
used to know at what version this support stopped in my head, I have 
forgotten, and so far have not found that with an online search.

Most of the C runtime routines that formerly required the POSIX 
subsystem are in the VMS C runtime, so there no separate POSIX builds 
needed for newer versions of VMS.

The behavior of the VMS C runtime is controlled by feature settings that 
can be set by logical names before the program is run.

These feature settings include the support of longer filenames with 
almost any characters including UTF-8, on ODS-5 volumes and if routines 
should behave in a more Unix like fashion.

I will be adding vms_crtl_init.c to the next commit for vmsbuild.com and 
descrip.mms.  This vms_crtl_init.c detects when the program is running 
under a bash or similar shell and sets the feature settings for better 
UNIX compatibility, or sets them for better DCL compatibilities.
It makes sure that the extended filename support is enabled.
This way a single binary can be used for both the VMS environment and 
the DCL environment.

As a replacement, a GNV https://sourceforge.net/projects/gnv/ was 
created.  There is older version of gawk that is provided by that package.

Recently a group of programmers have started trying to correct issues 
with the GNV project and reorganize it as a collection of packages along 
wth a VMSPORTS https://sourceforge.net/projects/vms-ports/ effort.

The GNV environment provides a Bash shell and several other utilities 
intended to simulate a Unix environment.

The web page https://sourceforge.net/p/gnv/wiki/InstallingGNVPackages/ 
documents what needs to be done before installing these newer packages 
while GNV is in the transition phase.

Currently only Bash 4.2.45 and Coreutils 8.21 have been packaged, both 
are fixing critical issues in GNV.

I plan to have a GAWK PCSI package that will be installed with the VMS 
PRODUCT command.  This package will install into the GNV directory 
structure, which mimics a Linux directory tree, so I will need to 
determine where to put the gawk.hlp file in it.

Regards,
-John



On 12/6/2013 7:20 PM, John E. Malmberg wrote:
> Hello Arnold,
>
> I almost missed this, the gmail randomizer threw it in the spam folder.
>   It tends to do that with about 1 to 5 percent of my incoming e-mail.
>
> I am still learning git, so I am not sure how to regenerate the patch
> set.  If you can help me with the commands, I can try to regenerate it.
>
> I ended up doing two patches, because I forgot that VMS execute
> permission is interpreted differently than Unix, so is usually on.
>
> I did not change the commands to kick off the build procedure, just the
> internals.
>
> Unfortunately  Texinfo is not in my toolbox.  I will look at the *.tex*
> files to see what I can figure out for editing the source format for any
> changes.  I did update the readme.vms FILE.
>
> Also this patch should enable the dynamic extension loading, which I
> forgot to put in the change list.
>
> To better support dynamic extensions, gawk on VMS/ALPHA should be built
> using IEEE floating point.
>
> This slightly changes the precision of floating point output.  I am not
> sure that anyone would notice.  IA64 defaults to IEEE, and VAX can not
> support IEEE.
>
> The reason for this is that most open source packages expect the IEEE
> behavior, so it seems best to default to that when possible.
>
> When we get this step resolved, I am going to work on the procedure to
> generate a installable package in the VMS PCSI format.
>
> Regards,
> -John
>
>
>
> On 12/6/2013 2:40 AM, arnold@skeeve.com wrote:
>> Hi.
>>
>> Thanks for this.  I already pushed the change you suggested about
>> version.c.  Do you want to pull and regenerate this patch set?
>> Or should I just apply it all and you'll send me an updated patch?
>>
>> Finally, we need the manual updated about the build procedure. If
>> Texinfo isn't in your toolbox, you can just send me plain text and I'll
>> handle the formatting, but the manual needs to be updated and also
>> and README files.
>>
>> This is great work and it's exciting that VMS gawk is progressing
>> this way.
>>
>> Thanks,
>>
>> Arnold
>>
>> "John E. Malmberg" <wb8tyw@qsl.net> wrote:
>>
>>> Forgot to document that the previous strftime test was incorrect as it
>>> was not specifying a timezone on VMS versions that support timezones
>>> which made the behavior different than on Unix/Linux.  The test has been
>>> fixed.
>>>
>>> All tests now passing on VAX/VMS 7.3, Alpha and IA64/VMS 8.4.
>>> Tests are still running on ALPHA/VMS 8.3, but I am expecting them to
>>> also pass.
>>>
>>> VMS build procedures no longer need to be edited when the version number
>>> changes.
>>>
>>> 2013-12-05   John E. Malmberg       <wb8tyw@qsl.net>
>>>
>>> * New config_h.com to generate config.h
>>>
>>> * New gawk_ident.com generates ident line for link option file.
>>>
>>> * Add version_c.com to create version.c from version.in.
>>>
>>> * Remove fcntl.h covering up real fcntl.h.  If an older version
>>>     of VMS needs this file, the build procedure should be updated
>>>     to generate it from a template.
>>>
>>> * descrip.mms: Use command files to generate files based
>>>     on same input files as a Linux build.
>>>
>>> * gawkmisc.vms (files_are_same): support _USE_STD_STAT for VMS 8.x.
>>>
>>> * generate_config_vms_h_gawk.com: Generates a helper file
>>>     config_vms.h to cover issues config_h.com can not handle.
>>>
>>> * vmsbuild.com: Use command files to generate files based
>>>     on the same input files as a Linux build.
>>>
>>> * vms_misc.c (vms_open): VMS CRTL setting errno to ENOENT where
>>>     it should be set to EMFILE.
>>>
>>> Regards,
>>> -John
>

From wb8tyw@gmail.com  Mon Dec 16 21:27:20 2013
Return-Path: <wb8tyw@gmail.com>
Received: from skeeve.com (skeeve.com [127.0.0.1])
	by skeeve.com (8.14.4/8.14.4/Debian-2ubuntu2.1) with ESMTP id rBGJQJfQ002202
	for <arnold@localhost>; Mon, 16 Dec 2013 21:27:19 +0200
X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on
	frenzy.freefriends.org
X-Spam-Level: 
X-Spam-Status: No, score=-2.7 required=5.0 tests=BAYES_00,DKIM_SIGNED,
	DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW autolearn=ham
	version=3.3.1
X-Envelope-From: wb8tyw@gmail.com
X-Envelope-To: <arnold@skeeve.com>
Received: from frenzy.freefriends.org [66.54.153.139]
	by skeeve.com with IMAP (fetchmail-6.3.21)
	for <arnold@localhost> (single-drop); Mon, 16 Dec 2013 21:27:19 +0200 (IST)
Received: from mail-ie0-f173.google.com (mail-ie0-f173.google.com [209.85.223.173])
	by freefriends.org (8.14.6/8.14.6) with ESMTP id rBG6Tatb002732
	for <arnold@skeeve.com>; Sun, 15 Dec 2013 23:29:39 -0700
Received: by mail-ie0-f173.google.com with SMTP id to1so5932042ieb.4
        for <arnold@skeeve.com>; Sun, 15 Dec 2013 22:29:36 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20120113;
        h=message-id:date:from:user-agent:mime-version:to:cc:subject
         :content-type;
        bh=ksiNjtwxRvh74kzY1LKaRL/l+Ja2PriUxgPwwm+Igho=;
        b=BiJLgaDXXU4R0IwBKeNvon0qca3q3Le4HtVArPhKruA9rRhSqfVcTRHrQ8fF1r0f5h
         N28vQidBEX3PNHi9k0uzfHF1sqT7LHGxOfSS/7ipk8UWoa4BVB3taSOFAAMzTkQUeduC
         2RNaWLJHjFR4xRvkOwYqJPVikLyg2atmds6iLoxKOFbNCz+KUFQacDmaCLiibEg4pxV1
         DsZWlBKLFrl8wXF5+gy3ZAj3FafIZluCgY5rUqrfXxrV9a7kumGrgljPG84MAOT01eCe
         nXnlrhJgxIAjndT7sYPSYGZkeieSagmN57kRqefNkHk9PhdF5RbY3J0PBC7r0cJn/MKE
         Uh0A==
X-Received: by 10.50.154.102 with SMTP id vn6mr13663789igb.1.1387175375953;
        Sun, 15 Dec 2013 22:29:35 -0800 (PST)
Received: from [192.168.0.101] (173-19-202-206.client.mchsi.com. [173.19.202.206])
        by mx.google.com with ESMTPSA id da14sm14704681igc.1.2013.12.15.22.29.33
        for <multiple recipients>
        (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128);
        Sun, 15 Dec 2013 22:29:34 -0800 (PST)
Message-ID: <52AE9DD7.1020409@gmail.com>
Date: Mon, 16 Dec 2013 00:29:43 -0600
From: John Malmberg <wb8tyw@gmail.com>
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:17.0) Gecko/20130215 Thunderbird/17.0.3
MIME-Version: 1.0
To: Aharon Robbins <arnold@skeeve.com>
CC: "John E. Malmberg" <wb8tyw@qsl.net>
Subject: Updates for gawktexi.in,readme.vms
Content-Type: multipart/mixed;
 boundary="------------040502020402090508010905"
Status: RO
X-Status: A

This is a multi-part message in MIME format.
--------------040502020402090508010905
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hello Arnold,

These are the preliminary updates for the VMS documentation.

I forgot to document (not sure where to put it)

   *  The exit code reporting in GAWK is wrong and violates VMS
      standards.  This is a common problem in older ports because
      the exit() was not implemented correctly, and neither was
      the rest of the C runtime.  Day 1 porting bug that was inevitable.

      The first problem is that if I fix it to report the correct exit
      status, it would break DCL procedures written to expect the
      wrong value.

      The second problem is that the new exit() call does not
      allow setting the severity values and the existing gawk code
      is setting them for tests to use.

      Using reverse engineering, I have determined how to encode
      the correct exit codes with severity into the old exit()
      API.

   If the shell is a UNIX shell:

   *  The input parameters and options are handled the same as Unix.
      (I have forgotten if I sent you that patch yet)

   *  The exit codes are correct.  The DCL severity information
      is also added even though most C programs will not use them.

I do not know if would be permissible to change the behavior to always 
use the correct exit code with the severity values set.

Existing DCL code using the severity values would not notice this change.

I would have to document how to convert UNIX exit codes to DCL values 
for people to get the original codes back.



In the documentation, I have been trying to change the case of the 
filenames referenced in the VMS sections to be exact instead in upper case.


In general, on the older ODS-2 VMS file system, the files were stored 
and displayed in upper case.

In the newer ODS-5 file system, filenames are stored in a case preserved 
state and are when unpacked from a repository are still in lower case.

The VMS DCL shell also has an optional mode /PARSE=EXTEND mode where it 
handles those names.

In the normal mode DCL converts all unquoted parameters to upper case, 
and then the C runtime converts all unquoted parameters to lower case.
A slight but significant difference.

In the extended mode, DCL and the C runtime will also not do any 
conversion of parameters.

Which means that in extended mode, the target to the MMS or MMK program 
needs to be in the exact case, or it may not be found.

When you have an NFS mounted volume, like I do, things get a bit strange 
on older VMS, like VAX/VMS, because it presents lower case filenames in 
upper case, and uses the $ character to indicate when the case gets 
inverted.

Example:

TOAD> dir src_root:[gawk.readme_d]readme.vms
%DIRECT-E-OPENIN, error opening SRC_ROOT:[GAWK.README_D]README.VMS;* as 
input

TOAD> dir src_root:[gawk.$readme_$d]$readme.vms

Directory SRC_ROOT:[GAWK.$README_$D]

$README.VMS;1

On current versions of VMS that support ODS-5, the exact case of the 
files are seen.

EAGLE> dir src_root:[gawk.readme_d]readme.vms

Directory SRC_ROOT:[gawk.README_d]

README.VMS;1

This affects build procedures as they must look for both path names.

Regards,
-John

--------------040502020402090508010905
Content-Type: text/plain; charset=windows-1252;
 name="gawktexti_in.gdiff"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="gawktexti_in.gdiff"

--- /src_root/gawk/doc/gawktexi.in	Fri Dec 13 17:43:41 2013
+++ /vms_root/gawk/doc/gawktexi.in	Sun Dec 15 23:28:13 2013
@@ -32713,6 +32713,9 @@
 @item
 Prestandard VAX C compiler for VAX/VMS
 
+@item
+GCC for VAX and Alpha has not been tested for a while.
+
 @end itemize
 
 @end itemize
@@ -33993,41 +33996,96 @@
 @appendixsubsubsec Compiling @command{gawk} on VMS
 @cindex compiling @command{gawk} for VMS
 
-To compile @command{gawk} under VMS, there is a @code{DCL} command procedure that
-issues all the necessary @code{CC} and @code{LINK} commands. There is
-also a @file{Makefile} for use with the @code{MMS} utility.  From the source
-directory, use either:
+To compile @command{gawk} under VMS, there is a @code{DCL} command procedure
+that issues all the necessary @code{CC} and @code{LINK} commands. There is
+also a @file{Makefile} for use with the @code{MMS} or @code{MMK} utility.
+From the source directory, use either:
 
 @example
-$ @kbd{@@[.VMS]VMSBUILD.COM}
+$ @kbd{@@[.vms]vmsbuild.com}
 @end example
 
 @noindent
 or:
 
 @example
-$ @kbd{MMS/DESCRIPTION=[.VMS]DESCRIP.MMS GAWK}
+$ @kbd{MMS/DESCRIPTION=[.vms]descrip.mms gawk}
 @end example
+or:
+
+@example
+$ @kbd{MMK/DESCRIPTION=[.vms]descrip.mms gawk}
+@end example
+
+The @code{MMK} is an open source freeware near clone of @code{MMS} and
+can better handle @code{ODS-5} volumes with upper and lower case filenames.
+@code{MMK} is available from @url{https://github.com/endlesssoftware/mmk}.
+
+With @CODE{ODS-5} volumes and extended parsing enabled, the case of the target
+parameter may need to be exact.
 
 Older versions of @command{gawk} could be built with VAX C or
 GNU C on VAX/VMS, as well as with DEC C, but that is no longer
 supported.  DEC C (also briefly known as ``Compaq C'' and now known
 as ``HP C,'' but referred to here as ``DEC C'') is required.  Both
-@code{VMSBUILD.COM} and @code{DESCRIP.MMS} contain some obsolete support
+@code{vmsbuild.com} and @code{descrip.mms} contain some obsolete support
 for the older compilers but are set up to use DEC C by default.
 
-@command{gawk} has been tested under Alpha/VMS 7.3-1 using Compaq C V6.4,
-and on Alpha/VMS 7.3, Alpha/VMS 7.3-2, and IA64/VMS 8.3.@footnote{The IA64
-architecture is also known as ``Itanium.''}
+@command{gawk} has been tested under VAX/VMS 7.3 and Alpha/VMS 7.3-1
+using Compaq C V6.4, and Alpha/VMS 7.3, Alpha/VMS 7.3-2, and IA64/VMS 8.3.
+Most recent builds are were using HP C V7.3 on Alpha VMS 8.3 and both
+Alpha and IA64 VMS 8.4 using HP C 7.3.
+@footnote{The IA64 architecture is also known as ``Itanium.''}
+
+Work is currently being done for a procedure to build @code{gawk} and create
+a PCSI kit for compatible with the GNV product.
+
+@appendixsubsubsec Compiling @command{gawk} Dynamic Extensions on VMS
+
+Dynamic extensions need to be compiled with the same compiler options for
+floating point, pointer size, and symbol name handling as @code{gawk}.
+Alpha and Itanium should use IEEE floating point.  The pointer size is 32 bits,
+and the symbol name handling is to be exact case with CRC shortening for
+symbols longer than 32 bits.
+
+Alpha and Itanium:
+
+@example
+/name=(as_is,short)
+/float=ieee/ieee_mode=denorm_results
+@end example
+
+VAX:
+
+@example
+/name=(as_is,short)
+@end example
+
+Compile time macros needed to be defined before the first VMS supplied
+header file is included.
+
+@example
+#if (__CRTL_VER >= 70200000) && !defined (__VAX)
+#define _LARGEFILE 1
+#endif
+
+#ifndef __VAX
+#ifdef __CRTL_VER
+#if __CRTL_VER >= 80200000
+#define _USE_STD_STAT 1
+#endif
+#endif
+#endif
+@end example
 
 @node VMS Installation Details
 @appendixsubsubsec Installing @command{gawk} on VMS
 
-To install @command{gawk}, all you need is a ``foreign'' command, which is
-a @code{DCL} symbol whose value begins with a dollar sign. For example:
+To use @command{gawk}, all you need is a ``foreign'' command, which is a
+@code{DCL} symbol whose value begins with a dollar sign. For example:
 
 @example
-$ @kbd{GAWK :== $disk1:[gnubin]GAWK}
+$ @kbd{GAWK :== $disk1:[gnubin]gawk}
 @end example
 
 @noindent
@@ -34039,10 +34097,15 @@
 @file{sylogin.com} procedure, which allows all users
 to run @command{gawk}.
 
+If your @command{gawk} was installed by a PCSI kit into the
+@file{GNV$GNU:} directory tree, the program will be known as
+@file{GNV$GNU:[bin]gnv$gawk.exe} and the help file will be
+@file{GNV$GNU:[vms_help]gawk.hlp}.
+
 Optionally, the help entry can be loaded into a VMS help library:
 
 @example
-$ @kbd{LIBRARY/HELP SYS$HELP:HELPLIB [.VMS]GAWK.HLP}
+$ @kbd{LIBRARY/HELP sys$help:helplib [.vms]gawk.hlp}
 @end example
 
 @noindent
@@ -34106,6 +34169,19 @@
 of @env{AWKPATH} is a comma-separated list of directory specifications.
 When defining it, the value should be quoted so that it retains a single
 translation and not a multitranslation @code{RMS} searchlist.
+
+@node VMS GNV
+
+The VMS GNV package provides a build environment similar to POSIX with ports
+of a collection of open source tools.  The @command{gawk} found in the GNV
+base kit is an older port.  Currently the GNV project is being reorganized
+to be individual PCSI packages for each component.
+@url{https://sourceforge.net/p/gnv/wiki/InstallingGNVPackages/}
+
+The normal build procedure @command{gawk} will produce a program that
+is suitable for use with GNV.  At this time work is being done to create
+the procedures for building a PCSI kit to replace the older @code{gawk}
+port.
 
 @ignore
 @c The VMS POSIX product, also known as POSIX for OpenVMS, is long defunct

--------------040502020402090508010905
Content-Type: text/plain; charset=windows-1252;
 name="readme_vms.gdiff"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="readme_vms.gdiff"

--- /src_root/gawk/readme_d/README.VMS	Fri Dec 13 17:43:41 2013
+++ /vms_root/gawk/readme_d/README.VMS	Sun Dec 15 22:57:42 2013
@@ -37,6 +37,38 @@
 GAWK was originally ported for VMS V4.6 and up.  It has not been tested
 with a release that old for some time.
 
+Compiling dynamic extensions on VMS:
+
+Dynamic extensions need to be compiled with the same compiler options for
+floating point, pointer size, and symbol name handling as gawk.
+Alpha and Itanium should use IEEE floating point.  The pointer size is 32 bits,
+and the symbol name handling is to be exact case with CRC shortening for
+symbols longer than 32 bits.
+
+Alpha and Itanium:
+
+/name=(as_is,short)
+/float=ieee/ieee_mode=denorm_results
+
+VAX:
+
+/name=(as_is,short)
+
+Compile time macros needed to be defined before the first VMS supplied
+header file is included.
+
+#if (__CRTL_VER >= 70200000) && !defined (__VAX)
+#define _LARGEFILE 1
+#endif
+
+#ifndef __VAX
+#ifdef __CRTL_VER
+#if __CRTL_VER >= 80200000
+#define _USE_STD_STAT 1
+#endif
+#endif
+#endif
+
 
 Installing GAWK on VMS:
 
@@ -47,6 +79,10 @@
 That symbol should be placed in the user's login.com or in the system-
 wide sylogin.com procedure so that it will be defined every time the
 user logs on.
+
+If your gawk was installed by a PCSI kit into the GNV$GNU: directory tree,
+the program will be known as GNV$GNU:[bin]gnv$gawk.exe and the help file
+will be GNV$GNU:[vms_help]gawk.hlp.
 
      Optionally, the help entry can be loaded into a VMS help library.
  |$ LIBRARY/HELP SYS$HELP:HELPLIB [.VMS]GAWK.HLP

--------------040502020402090508010905--

From wb8tyw@qsl.net  Wed Dec 11 20:03:44 2013
Return-Path: <wb8tyw@qsl.net>
Received: from skeeve.com (skeeve.com [127.0.0.1])
	by skeeve.com (8.14.4/8.14.4/Debian-2ubuntu2.1) with ESMTP id rBBI3gNX002464
	for <arnold@localhost>; Wed, 11 Dec 2013 20:03:43 +0200
X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on
	frenzy.freefriends.org
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM
	autolearn=ham version=3.3.1
X-Envelope-From: wb8tyw@qsl.net
X-Envelope-To: <arnold@skeeve.com>
Received: from frenzy.freefriends.org [66.54.153.139]
	by skeeve.com with IMAP (fetchmail-6.3.21)
	for <arnold@localhost> (single-drop); Wed, 11 Dec 2013 20:03:43 +0200 (IST)
Received: from Encompasserve.org (Eisner.encompasserve.org [67.53.90.116])
	by freefriends.org (8.14.6/8.14.6) with ESMTP id rBB5jaW9020988
	for <arnold@skeeve.com>; Tue, 10 Dec 2013 22:45:40 -0700
Received: from [192.168.0.101]
 (173-19-202-206.client.mchsi.com [173.19.202.206])
 by Encompasserve.org (PMDF V6.6 #13031)
 with ESMTPSA id <01P1SYL5LELG002FZE@Encompasserve.org> for arnold@skeeve.com;
 Tue, 10 Dec 2013 23:45:35 -0600 (CST)
Date: Tue, 10 Dec 2013 23:45:36 -0600
From: "John E. Malmberg" <wb8tyw@qsl.net>
Subject: Re: Some questions...
In-reply-to: <201312110426.rBB4QXXc003297@skeeve.com>
To: Aharon Robbins <arnold@skeeve.com>
Cc: wb8tyw@qsl.net
Message-id: <52A7FC00.2040905@qsl.net>
MIME-version: 1.0
Content-type: text/plain; charset=ISO-8859-1; format=flowed
Content-transfer-encoding: 7BIT
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:17.0) Gecko/20130215
 Thunderbird/17.0.3
References: <529A9537.7090404@qsl.net> <201312110426.rBB4QXXc003297@skeeve.com>
Status: R

On 12/10/2013 10:26 PM, Aharon Robbins wrote:
> Hi. Some questions about your changes.
>
> 1. I notice that there is at least one instance of #if __VMS, where as
> it seems most code uses #if VMS.  Is that on purpose?  If not, which
> is correct?

VMS is a macro historically pre-defined by C compilers on C before the 
ANSI rules.

Under strict ANSI rules, a standards compliant compiler is not allowed 
to pre-define macros that are not prefixed by "__".

Which means that you can not count on the VMS macro being defined unless 
you do it yourself.

So __VMS is the ANSI correct definition.  But it would be a lot of work 
to change all the code.

The vms_config.h that is now generated has these lines in it to make 
sure that VMS is defined.

#ifdef __VMS  /* ANSI compliant */
#ifndef VMS /* PRE-ansi */
#define VMS 1
#endif
#endif

This may be needed for other architectures.  It only shows up as an 
issue if you set a "strict" compliance flag on the compile command.

> 2. For the include of the main wrapper.  Gawk already has a routine
> os_arg_fixup which is intended to manipulate argc and argv as needed
> by an OS-specific routine before main() parses it.  A quick look at
> the wrapper looks to me like it could be made to fit within that framework,
> which would keep #if and the include out of the mainline code.

I can look at that.  I can also look to see if I can move the logic from 
vms_crtl_init.c into it in a way that would sidestep that it has no 
copyright notice at all.  The result would be code that is almost 
identical to code that is in the official Perl distribution.

GNV was started as a user project and when the VMS POSIX product was 
dropped and they needed a Unix like environment, Digital took over the 
GNV product, but did not really do much to keep it up to date.

The routine in vms_crtl_init.c is actually called before main() is 
called, so just linking a module with it in is all that it takes to 
activate the change.  Unfortunately I do not know how to get it to 
magically fix up the argv array.

> If that looks reasonable to you, can you make that change?  You can send
> it as a diff against what you currently have already done.

I will look at it tomorrow night.  I also found another change needed 
for vms_args.c to work properly under a bash shell.

> I'm sorry about the hassle on the copyrights, but I really have to play
> this game by the FSF rules.  I hope we can get it clarified soon
> so that I can include the files in the dist.

I hope so.  I have not seen a reply from Karl yet.  He did say that 
there are other options.

This would clear the way for getting the similar code accepted into the 
Bash and coreutils repositories.  I have a report that the first bash 
4.3 beta built on VMS and looked real good on the test run.

I am currently working on adapting the packaging scripts from bash and 
coreutils to gawk.  Each product has had it unique features, so I have 
not yet come up with a generic set of scripts.

This is a preview what will be coming:

backup_gawk_src.com - Builds a VMS equivalent to tar archive.  By 
placing the source in the PCSI kit, it covers the requirement of 
providing source somewhere with the binary.

build_gawk_pcsi_desc.com - Creates a VMS PCSI package manifest file.

build_gawk_pcsi_text.com - text output by the VMS PCSI package installer.

compare_gawk_source.com - I have the source checked out on an NFS 
server.  The backup program needs me to copy it to a VMS volume first. 
This procedure makes sure that the two copies are the same and ready for 
backup.

gawk_alias_setup.com - Sets up gawk and awk links to gnv$gawk.exe. 
Needed to repair when sins of the past partially break an install.

gawk_verb.cld - for gawk to use gnv$gawk.exe as an image.  Different 
than gawk.cld.  To properly add an image to the VMS command table, it 
needs a prefix.  I have "GNV$" registered as a prefix for this purpose.

gnv_gawk_kit_name.com - Calculates the VMS PCSI package name, used for 
both file names and file contents.

gnv_gawk_startup.com - Run at VMS startup to make sure that the 
environment is set up properly for gawk.

pcsi_gawk_file_list.txt - Source file for build_gawk_pcsi_desc.com.

remove_old_gawk.com - Needed to properly clean up sins of the past. 
Only touches the the GNV provided gawk and awk programs.

stage_gawk_install.com - Does a pre-install into a fake "root" directory 
tree for the kitting procedure to pull files out of.

Regards,
-John