Talk:Rexx

Recent edit

A recent edit changed case insensitive variables to typeless variables. There were two problems with that:

1. The article already listed dynamic data typing (no declarations)
2. No replacement text was provided for case independence.

I am reverting the change and then replacing case insensitive variables with case independent tokens, including variable names.

In the same edit I'm including VM and TSO/E as bundled with REXX and I'm mentioning that there are potential surprises for PL/I programmers. Shmuel (Seymour J.) Metz Username:Chatul (talk) 00:36, 12 September 2010 (UTC)

Rexx is not Object Rexx

The infobox shows Rexx as object oriented, which it is not. There is a separate article on Object REXX, where that paradign does apply. Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:33, 31 December 2010 (UTC)

NPOV: NOVALUE

The reference to NOVALUE violates WP:NPOV; quoting feature is totally out of bounds and the disdain for allowing bare words to default to their upper case names is hardly universal. The text should be rewritten in a neutral fashion, although it is certainly appropriate for the text to describe the controversy between the convenience and safety camps. Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:23, 31 December 2010 (UTC)

Addressed. RossPatterson (talk) 02:01, 11 June 2014 (UTC)

The article still contains the sentence Virtually all serious Rexx programs contain signal on novalue or a similar statement., which violates WP:NPOV. Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:57, 12 June 2014 (UTC)

Feel free to change it. RossPatterson (talk) 22:17, 12 June 2014 (UTC)

Reworded without "many" WP:WEASEL. –2A03:2267:0:0:452A:FA00:6D62:CEE8 (talk) 11:35, 10 January 2017 (UTC)

First line

The first line of a Rexx script does not identify the operating system; it identifies the compiler or interpreter. There are several different uses of a special first line.

• The presence of a comment normally identifies the script as Rexx
• In TSO, a Rexx script loaded from SYSPROC must start with /* REXX */ to distinguish it from a clist
• In Unix the first line may be a shebang to identify the language processor
• In OS/2 the first line may be EXTPROC to identify the language processor; normally that is not necessary as an initial comment will let the script run with the default Rexx.

Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:54, 10 November 2011 (UTC)

Community development

I think that many IBM people would say that REXX's design was highly developed by a community of IBM personnel and customers I think. This was before the internet was in use but PROFS provided email capability at the time. The fact that it was a community-designed language is noteworthy but I cannot find a reference. I hope an IBM person can. Sam Tomato (talk) 17:40, 9 June 2014 (UTC)

I don't know of any community development of Rexx per se, but here was massive community development of function packages and Rexx-aware commands to support applications written in Rexx, to say nothing of such development tools as VX-REXX and VisPro REXX. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:12, 10 June 2014 (UTC)

I would assume Sam Tomato is referring to the process Mike Cowlishaw used for having the REXX user community inside IBM vett new features and substantive changes. He has spoken of this many times, and remarked on how valuable it was for developing the language. See, for example, p. 334 in Mike's 1984 IBM Systems Journal article The design of the REXX language:

The most important factor in the development of REXX began to take effect when the first interpreter was distributed over the IBM communication network known as VNET. (This network links over 1400 mainframe computers in forty countries.) From the beginning, many hundreds of people were using the language. All these users, from temporary staff to professional programmers, were able to provide immediate feedback to the designer on their preferences, needs, and suggestions for change. An informal language committee then appeared spontaneously and communicated among themselves and with the designer entirely electronically. The discussions of the committee grew to be hundreds of thousands of lines, and these and the similar quantity of mail from the users were all kept for later review.

As time passed, it became clear that changes in the language were necessary. Using the network, the designer could interactively explain and discuss the changes that were required, some of which were incompatible with the then-current version of the language. The decision to make an incompatible change was never taken lightly, but-because changes could be made relatively easily and explained to users in detail-the language was able to evolve much further than would have been the case if upward compatibility only were considered. Several other important concepts guided the process of enhancing the language.

RossPatterson (talk) 22:47, 10 June 2014 (UTC)

It's a wiki, just add the info where it fits to the article, I think I could add a TRL reference, or just use the link shown above, or a link to this document archived on Mike's web site. VX-REXX was 15 (?) years later, that's something different. –Be..anyone (talk) 14:32, 13 January 2015 (UTC)

Strange remark

To me 'able to fully recover following a fatal error' is contradictory: if a full recovery is possible then the error was not fatal. — Preceding unsigned comment added by 86.185.216.95 (talk) 21:30, 4 November 2014 (UTC)

It seems clear in context that here fatal means that it would terminated execution if not intercepted. If you believe that it is clearer, you could change it to following an otherwise fatal error, but I believe that it is clear as is. Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:47, 6 November 2014 (UTC)

Ten years later... I agree. It's nonsense to talk about recovering from a fatal error. Fatal error has a (generally accepted) meaning: an error from which there is no recovering! I plan to change to "... an otherwise fatal error" Stevebroshar (talk) 18:26, 12 January 2025 (UTC)

Missing key concepts

It is difficult to appreciate the flavor of Rexx as a scripting language without an understanding of the Rexx variable pool and the environment, especially the definition of function packages related to a particular environment. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:41, 23 February 2015 (UTC)

1. References and notes on talk pages are gross.
2. REXX is dead, only very old folks like you and me recall what it was and still use it, everybody else uses Python or Lua.
3. IIRC SHVENV is no official part of the language, maybe the stack would come first as missing concept.
4. But when you add this, that, or both it's of course your decision, one of the few nice features on a wiki.

–Be..anyone (talk) 05:12, 24 February 2015 (UTC)

(Restored my comment as it was, adding tags below what your replies were about: Feel free to remove that, but keep my list together, please. –Be..anyone (talk) 06:37, 25 February 2015 (UTC))

ad 1

You're entitled to your opinion; I find the information useful when present. Shmuel (Seymour J.) Metz Username:Chatul (talk) 23:30, 24 February 2015 (UTC)

ad 2

The corpse is under active development. Shmuel (Seymour J.) Metz Username:Chatul (talk) 23:30, 24 February 2015 (UTC)

ad 3

From ANSI® X3J18-199X American National Standard for Information Systems – Programming Language REXX

5.13 Variable pool
The variable pool interface consists of functions which the configuration shall provide to manipulate the variables and to obtain some characteristics of a REXX program.

These functions can be called from programs not written in REXX — commands and external routines invoked from a REXX program, or traps invoked from the language processor.

Shmuel (Seymour J.) Metz Username:Chatul (talk) 23:30, 24 February 2015 (UTC)

It's a wiki, as long as it passes WP:42 with a reference just do what you like. Claiming to speak for be..anyone: 2A03:2267:0:0:452A:FA00:6D62:CEE8 (talk) 11:42, 10 January 2017 (UTC)

File extionsions - types

The term file extension is problematical

1. Rex is based on the older, cognate, concept of file type, used in, e.g. CMS, TSO.
2. Rex also uses file types to identify Rex macros in Rex-aware applications, e.g., XEDIT. Note that this usage carries over to PC-based operating systems, e.,g., PC-DOS, Linux.

Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:06, 8 July 2015 (UTC)

For the purposes of enwiki we just need a .rex => REXX as noted on .rex, the fine print of "type" vs. "extension" should be handled elsewhere, it's no REXX-issue. PC DOS used .bat for the scripts and .rx for its REXXUTIL-subset, Regina also uses .rexx and .cmd, the latter is slightly confusing on Windows, ooRexx might know .orx and certainly supports .rexg (GUI) and .rexp (CLI+pause), and I find these details while looking at the PATHEXT environment value on my Windows box. –2A03:2267:0:0:452A:FA00:6D62:CEE8 (talk) 11:58, 10 January 2017 (UTC)

Degree of use?

A recent edit challenged the claim that Rexx is widely used. It is ubiquitous on IBM mainframes, and there are open source interpreters for the Intel world, so the claim is plausible. Does anybody have any hard data on the usage prevalence of, e.g., OOREXX, Regina, in *bsd, Linux or windows? Shmuel (Seymour J.) Metz Username:Chatul (talk) 23:31, 6 February 2017 (UTC)

From 2005 " Rexx's worldwide user community numbers in the hundreds of thousands" [1] Peter Flass (talk) 03:22, 7 February 2017 (UTC)

Appropriate chronology for citations?

@Polluks: A recent edit added a citation to "VM/SP R3" that is for VM/ESA V2 R4.0. Wouldn't it be more appropriate to cite either the manual for the most recent z/VM version or for the original VM/SP R3 version? Should "TSO/E V2" also have a citation? Also, do the citations belong in the infobox or should they be in the article proper? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 11:24, 26 May 2023 (UTC)

Indeed. Polluks ★ 16:47, 28 May 2023 (UTC)

@Polluks: Move citations out of infobox, use citations for original releases, or both? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 12:58, 29 May 2023 (UTC)

I've update the infobox to use the oldest sources I could find for VM, TSO and SAA REXX. The most recent VM sources can be found at z/VM Library Overview.

References

promotional edits for RexxUtil

Using "best known", etc., is promotional editing rather than following the Wikipedia guidelines WP:NPOV TEDickey (talk) 10:13, 15 October 2023 (UTC)

I don't agree that it is promotional, but I also don't believe that popularity ratings of packages or language features, even if cited, are helpful in general, so unless there is some reason that the reader would care ... -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:35, 16 October 2023 (UTC)

"Best known" may not be promotional, but it is subjective, disputable, opinion ... which I think is what WP guidelines mean by point of view. I often remove such language but if I leave it in I add wiggle phrases like some contend or think or arguably ... Stevebroshar (talk) 10:41, 14 January 2025 (UTC)

RexxInfo external link removed by user Tedicky

@Tedicky I do not agree that adding an external link to rexxinf.org violates anything in WP:EL. I think a referral to a wikipedia rule alone does not sufficiently justify a revert of my addition. The linked to website is an important resource for any Rexx user; more so, for example, than the link to OpenHub above it. Please revert your revert. René Vincent Jansen (talk) 11:11, 21 April 2024 (UTC)

Ok, looking at it now, I agree. René Vincent Jansen (talk) 10:14, 22 April 2024 (UTC)

A programming language that can be interpreted or compiled

WRT "is a programming language that can be interpreted or compiled." Adds no value. All languages can be interpreted or compiled! A language defines syntax and semantics; not how the source code gets translated into a runnable form. Some languages are typically compiled (C++) and some are typically interpreted (Python) but C++ is not inherently characterized as being compiled nor is Python inherently interpreted. We could write a compiler for python and an interpreter for C++. That no one has is interesting but does not change the nature of the languages. Some languages are typically compiled to byte code (c#, java). But is it really compilation if it's not to machine code? IDK. What about a translator that converts Ada to C++? Is that a compiler? IDK. Thing is, it doesn't matter. Languages (just) specify syntax and semantics. Translators allow a language to be useful. Stevebroshar (talk) 18:53, 12 January 2025 (UTC)

Typeless? or no?

WRT "Variables are typeless ... Thus a variable's type can vary with its use in the program" is contradictory. Either they have type or not. Some languages provide automatic typing. Like if you assign like "v=3" then it's a number, but if you assign like "s='asdf'" then it's a string. Is Rexx like that? Or are all variables typeless? or are they strings? Stevebroshar (talk) 12:40, 14 January 2025 (UTC)

Notes

I did convert some info formatted as notes into normal body text. IMO notes should only be used in very rare cases. Normally, they make the info harder to read and harder to maintain. Further, if the info is not important enough to be in the body of the text, then it's probably not important enough to be in the article at all (even as a note). If the reason for using a note is bc organizing information is challenging; hard to make it flow/readable, then the solution is to figure out how to write it so the info does flow. It's a writing thing. Stevebroshar (talk) 14:30, 14 January 2025 (UTC)

Clarification of EXEC versus EXEC2

In z/VM, a script name foo EXEC bar may be in one of three different languages, EXEC, EXEC 2 or REXX.

If the first line contains a REXX comment, then the script is REXX.

If the first line is an &TRACE statement then the script is EXEC2.

Otherwise the script is EXEC.

How much of this, if any, belongs in the article? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:17, 14 January 2025 (UTC)

Good question. Without citation it appears to be OR, but if it's correct I think we let someone else deal with citation. I trust that you know what you are saying ;o) But, my wife tells me I'm overly trusting. So there's that.

I think the detail is good and relevant. But, please avoid notes. They make info harder to read and to maintain. ... I don't agree with everything I learned in college, but I do think my writing professor was right about this point of style: If it's not important enough to include in the body of the text, it's probably not important enough to include as a note. Either put the info in the body of the text or drop it.

Going further, this article has lots of good info in it. But, I edited it extensively since it was poorly written. It seemed to target folks that already know the topic; or at least skilled programmers. I firmly believe that a WP article can have great content and be written well, that technical info is hard to write about, that all writing is hard. So, good luck. You can write about a highly technical topic in a way that the general public can consume. Stevebroshar (talk) 13:08, 15 January 2025 (UTC)

Notes

References

Types of statement

When new content is added to an article I ask myself: is this a valuable addition? ... I wasn't wondering how many types of statements rexx has, and I don't think knowing there are three is interesting. That is my opinion of course.

I do wonder whether this section is OR. Are there really three according to the language? Can we find a citation?

But... if we are going to add this, I'd say that each needs an example (even though examples are OR). Saying that a language supports assignment statement without an example assumes the reader knows that an assignment statement looks like. You know, I know, but does the non-programmer? the high school student? WP is not only for programmers.

Further, please limit use of comments in code examples. I'd say that is a programmer talking to a programmer. Minimize code! If you need to note something, put that in the text instead of a code comment. Stevebroshar (talk) 13:22, 15 January 2025 (UTC)

I can certainly find a citation, but if anybody knows of a secondary source that would be better, at least in the wiki context. Should the citation include |quote=?

Are you suggesting an example of each type? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:31, 15 January 2025 (UTC)

References

Case of keywords

I changed many of the language keywords that were in upper case to lower. Before my change, it was a mixed bag; many upper and many lower. I was trying to apply some consistency. I thought it odd that they were lower case in the example code (somewhat consistently) but upper case elsewhere (sometimes). But, some are still upper after my changes. I left NOP and the conditions as upper ... since I lacked confidence that changing them was a good idea. I think I made a good change and I think maybe the rest should be lower cased to match.

FWIW, I think maybe the keywords were in upper case to highlight them for the rest of the text. This might have been a common style a long time ago that has hung on for this context since it comes from along time ago. But, with markup (<code>) we can highlight without resorting to upper case. Stevebroshar (talk) 13:34, 15 January 2025 (UTC)

I lowercased nop and the conditions. Stevebroshar (talk) 13:56, 15 January 2025 (UTC)

No reserved keywords?

WRT "No reserved keywords" Seems that it has lots of them: do, end, while, until, index, forever, if, then, else...

Or can you define a variable with any of these words? If so, that should be noted (since it seems important and a horrible idea for a language!) Stevebroshar (talk) 13:55, 15 January 2025 (UTC)

What does numeric do?

The numeric section currently just shows examples. But I don't know what the examples tell me. What does numeric do? Stevebroshar (talk) 14:07, 15 January 2025 (UTC)

Organization

Most of the material in § Syntax relates to semantics rather than to syntax. I propose something like

• Syntax
  • Clauses
    • Null clause
    • Labels
    • Instructions
      • Assignments
      • Keyword Instructions
      • Commands:
• Assignments and variables
• Expressions
• Assignment clauses
• Keyword instruction clauses
• Commands to external environments
  • Address instruction
  • Command clauses

The actual headings should match the terms in the documentation adjusted to avoid duplicate anchors.

Is there guidance on synthetic examples versus exaamples copied from sources, to what extent fair uses allows copying examples, and how to properly attribute copied examples? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:01, 15 January 2025 (UTC)

I do not know whether there is WP guidance on examples. I doubt it. I do have opinions on the matter and I'll share them with you ... even though unsolicited ;) This is probably more than you're asking about, but I'm going to share any way ;)

I guess by "synthetic" you mean invented or something. I think that most programming examples on WP are invented by the article author. I think some authors include example code from their professional work and that many are _way_ too complicated to get the point across. Some examples probably come from reference material. For example(!), the example hello-world for C is basically a meme :) This is all just my observation & conjecture.

I said elsewhere that examples (invented/synthetic) are OR. But after some noodling I think that assertion is debatable. Considering that WP guidance is to avoid quotation, but to paraphrase, maybe synthetic/invented example code is essentially paraphrasing a language rule.

Some general thoughts on code examples:

• A good example is short and to the point. Avoid edge cases that explain something useful but off topic since that confuses the reader. Keep it simple (KISS).
• Code comments should be avoided since it makes it harder to read. Code comments are for coding; we're not coding. Be as gentle to the reader as possible. They might not be a programmer or might not know what constitutes a comment in a particular language. Comment in the prose; not in the code.
• Don't conflate syntax with code examples. This article, for example, has both syntax description blocks as well as code examples that IMO were not sufficiently different for the reader to notice a difference, so I modified the syntax blocks to look different. Maybe examples are better than syntax descriptions since the de facto syntax syntax (!) although known by many programmers is not well-defined and not known by non-programmers. I didn't replace the syntax blocks of this article with examples since I don't know this language and didn't want to invent examples that were wrong. (Notice how this implies that examples are OR)

Thoughts on your proposal:

• Prefer singular over plural (in general). It's more formal and encyclopedic
• Avoid "and" in a header. A header should be one thing; not multiple
• Flatter is better. Avoid hierarchal info as it's harder to understand than flat. I'm guessing you are a technical person since they often organize hierarchically ... since this sort of thinking is useful when breaking down a problem. But for writing in general think flatter.
• I agree that the section titled 'syntax' covers more than syntax. But let's not throw the baby out with the bath water. I think the current content and organization is pretty good and suggest keeping mostly as-is. We are not writing a language guide. We're explaining interesting aspects of Rexx to the world. We don't even have to be complete. We only need to hit the highlights. Prioritize engagement with the reader than to include every detail. I like how the section is organized by aspects that are approachable. For example, your proposal mentions clauses and expressions alot, but those are less approachable concepts for the lay person. We need to discuss them, but we should lead with the most approachable aspects.

As for a new section name: that's a hard one. Naming is hard! The section is about what the language provides and how it works. It's about syntax and semantics. It's about language definition. Could call it "Language", but one might say the whole article is about the language ... although it's actually about more than a language; for example, the interpreters and compilers are outside the language. Maybe "Features" ... but distinct from the current "Attributes" section which are high-level, touchy-feely. "Language features", "Language definition" ... I like that one. It's singular and clear and distinct from other section titles. "Definition" seems good; maybe better. No need to say language in an article about a language.

In general, I think this article is well organized. As you may have noticed I did re-organize it a little recently. But not much. Mostly I combined sections that had duplicate info and moved sections around for better flow.

In general, I think the content of the article is pretty good. ... except for the issues for which I added talk topics. The article has some holes and some contradictions that should be resolved. I'm considering buying Mike's book to look for answers. Stevebroshar (talk) 12:38, 17 January 2025 (UTC)

Yes, I am a Techie. I was trying to use the vocabulary form the ANSI standard and Mike's book. If you have a background in technical writing then I'll defer to your judgement as to whether that is desirable.

One important area that belongs in the article is the concept of environment, including the processing of commands and the keyword instruction ADDRESS. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:32, 17 January 2025 (UTC)

References

Was IBM involved with the development or not?

The intro says "Rexx ... developed at IBM" and in history it says "On his own time, Mike Cowlishaw developed the language..." I did edit these statements recently but I think this is essentially what both statements meant even before editing. As I read them today, they are contradictory. Either he did it as part of his work at IBM and therefore the company has some claim to Rexx or he did it on his own time while also working at IBM but the company has no claim to Rexx. Stevebroshar (talk) 11:26, 17 January 2025 (UTC)

On his own time but using IBM facilities. It's quite common for large organizations to allow or even encourage that. I don't know the exact sequence of events, but there was at least one official publication while it was still called REX. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 15:49, 17 January 2025 (UTC)

!-- Keep after last sig -->

References