The ANSI X3J18 Standard for Rexx today


Two years ago Rexx joined the elite group of about 20 programming languages for which there is an American National Standard.  The language level the Standard describes is the one in use then, "classic" Rexx before the time of Object Rexx and NetRexx, plus some extensions that the defining committee felt were necessary.  Maintenance of the "Spirit of Rexx" was ensured by Mike Cowlishaw's participation in all the meetings of the standardizing committee.  The RexxLA web site describes how the Standard was defined and how to get a copy.

Conformance to the standard is a matter for the developer's self-assessment; there is no conformance test suite.  As far as I know, Neil Milsted's implementation for UNIX is the only one to claim conformance, and that is not full conformance - the error messages are different from those the Standard specifies.  It seems that rather than investing in complete conformance, implementors have chosen to use the Standard as a criteria for smaller decisions.  In particular, the DATE and TIME conversions in many implementations are the same because they all chose to follow the Standard when adding these extensions to their implementations.

Since publication of the Standard the committee has continued to meet, combining the roles of potentially developing Object Rexx and NetRexx standards and of being the authority on language for current extensions. (For example, some implementations now allow the fourth and fifth arguments on DATE.  This language was not part of the Standard but is part of a potential new standard the committee is working on.)  Writing a formal technical proposal for defining the newer versions of Rexx is challenging, but practical.  Establishing such a proposal as a Standard may not be practical - the breadth of developer support is not what it was for "Classic" Rexx.

The process of approving an American National Standard involves a period when comments from the user community must be responded to.  In the case of Rexx, it is likely that those with valuable comments to make will mostly be members of RexxLA.  There is some discussion whether the cost of using American National Standard processes as a vehicle for future standardization is a best use of resources.

To get involved with this question (as well as a lot of grunt technical work) anyone is welcome to the meetings of the committee.  There is always one co-located with the Rexx Symposium, and there are others; the next is in Dallas, November 9-11.

Dr. Brian Marks
Chairman of the ANSI X3J18 Standards Committee for Rexx