NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/lang/rexx/739

Text File | 1992-08-19 | 5.2 KB | 128 lines

Newsgroups: comp.lang.rexx
Path: sparky!uunet!wrkgrp!ets
From: [e-mail] (Edward T Spire)
Subject: Re: Capturing System Command Output (was: SH Backquote)
Message-ID: <[e-mail]>
Organization: The Workstation Group
References: <[e-mail]>
Date: Wed, 19 Aug 92 19:15:42 GMT
Lines: 117

[e-mail] (Anders Christensen) writes:
: In article <[e-mail]> [e-mail] (Steve Bacher) writes:
:
: > Is anyone working on defining a standard syntax for this capability?
:
: If there anyone from the ANSI committee present, that would tell what
: they intend to do with this problem?

This area was discussed at the last session, and will be discussed again
next month.  Here's what was published at the end of the last session
on this issue.



(The following is an excerpt from ANSI document number X3J18/92-009,
"X3J18 REXX Standards Committee fourth meeting's discussion notes")



Commands to an External Environment

Two issues were discussed concerning commands to the external environment:
Portability of completion analysis, and standardization of the
redirection of the external command's stdin and stdout.


Portability of completion analysis.

Even though most implementations utilize a numeric RC with fairly
common standards (i.e., 0=normal completion) it was recognized that
the language definition cannot standardize the RC returned from the
external environment, therefore portability in analyzing the result
of external commands is a problem.

In fact, it was noted that some implementations allow the completion
status of the external command to be indicated independently from the
return code.

The committee is leaning towards providing yet another special variable
to indicate the completion status of the command to the external
environment in a standard manner.  We have some concern regarding
adding yet another special variable, so our first choice was to
use "result", which is already a special variable, which is not set
by an external command (according to TRL).  However, the AmigaDos
implementation of REXX (AREXX) does set "result" in response to some
external commands, and there was also concern that some programs
already using result would expect it to be preserved across an
external command.  Hence the committee is leaning towards adding the
special variable "rs", which would be assigned the value -1, 0, or 1
upon the completion of an external command, meaning failure, normal,
or error status.

We should note once again that the issue of removing all special variables
and replacing them with functions was once again raised as a way to
simplify the language, rather than further complicate it by adding
still more special variables.  An alternate suggestion of using a reserved
name space for special variables (i.e., _rc, _result, _rs, etc.) was
also discussed.


Standardization of the redirection of the external command's stdin and
stdout.

While many operating systems provide a means to redirect the standard
input and output streams for external commands to various sources,
there is little consistency among environments.  Often REXX language
extensions are required for various environments (such as the creation
of an external data queue for those environments that do not implement
one, including a mechanism to redirect command output there.)  Without
guidance from the language definition, such language extensions tend to
vary widely from platform to platform.

The user organizations among the committee members felt very strongly
that the standard should address this issue.  The committee settled on
a tentative syntax extension to the ADDRESS instruction that would
implement a redirection of the external command's stdin and/or stdout
I/O stream.  One possible syntax would be:

    ADDRESS environment [FROM source] [TO destination] [exprc]
      or
    ADDRESS [VALUE] exprv [FROM source] [TO destination]

The committee agreed that "source" could be one of:

   STDOUT (i.e., the REXX language processor's stdout stream)
   QUEUE

and "destination" could be one of:

   STDIN
   QUEUE (FIFO to the stack)
   PUSH  (LIFO to the stack.)

It was recognized that the connection of the external commands stdin/out
to the REXX program's stdin/out may be the less useful of the pair,
since it would interfere with the REXX program's normal use of stdin/out.

The user representatives were adamant that possible specifications for
"source" and "destination" should also include:

  VAR symbol
  STEM symbol.

to allow redirection to REXX variables.  The committee could not agree
that these latter extensions were warranted for the initial standard.
The former has line separation difficulties (what can be chosen as a
uniformly applicable line delimiter?), while the latter introduces a
new means of utilizing stem variables into the language, which should
be approached with some caution.  These last two options will certainly
be considered for inclusion in the second standard.

==========================================================================

Ed Spire                           email: [e-mail]      (on uunet)
The Workstation Group              voice: 800-228-0255
6300 River Road, Suite 700            or  708-696-4800
Rosemont, Illinois  60018            fax: [phone]