And some more background .. if I recall correctly, this was set (corrected) to work this way after an APAR (one of only two, ever) was raised for exactly this case because the interpreter did NOT report it as an error whereas the documentation implied that it should. There was some discussion in 1981 as to whether there should be a minimum value for NUMERIC DIGITS (e.g., 3), but it was hard to argue why that should be -- this would have been a cogent argument if we'd thought of this case! :-) (Separately, the original REX error messages .. the 'headline' error messages ... had to be really short because of the need to keep the interpreter within 32KB (yes, KB) so that it would fit in less then half of one rotation of paging drums ...) Mike
_____ From: Rick McGuire [mailto:object.r...@gmail.com] Sent: 09 November 2020 13:47 To: Open Object Rexx Developer Mailing List Subject: Re: [Oorexx-devel] Quirk of the Day On Mon, Nov 9, 2020 at 8:21 AM rvjan...@xs4all.nl <rvjan...@xs4all.nl> wrote: Hi Walter, Yes, well, it could be a typo. I am worried about all other implementations, including NetRexx and z/OS TSO, z/VM, Regina, brexx, etc, being wrong if this is the right way. It certainly fails the (however subjective) principle of least astonishment, and I also get the feeling the ‘numeric digits’ statement is not necessarily meant for the next instance of ‘numeric digits’. Changing the error message would go a long way: ‘with numeric digits set to X, Y is not a valid positive whole number.’ The mainframe versions all behave the same way. I have had this conversation several times since 1982, almost always from a tester playing with setting digits to 1. I’ll put it on the list for the ARB. Who wants to be on the ARB? We have an obligation to run an Architecture Review Board, as discussed during the symposium. I am looking for volunteers. I suggest at least Erich and Rick be on it. I suggest we do not convene more than once a quarter. René. On 9 Nov 2020, at 13:59, WalterPachl via Oorexx-devel <oorexx-devel@lists.sourceforge.net> wrote: Who on earth (except "us nasty testers") would ever issue ND 1???? I am much more concerned about the performance disaster I reported 2 or 3 weeks ago :-( Greetings WALTER "Rony G. Flatscher" <rony.flatsc...@wu.ac.at> hat am 9. November 2020 um 11:31 geschrieben: On 09.11.2020 10:35, Erich Steinböck wrote: This one stunned me! ~~~ numeric digits 1 numeric digits 18 -- Error 26.5: DIGITS value must be a positive whole number; found "18". ~~~ Maybe enhancing the error message to indicate the current setting of numeric digits which makes "18" not a positive whole number would explain to the programmer why the error occurs. Something like "Error 26.5: DIGITS value must be a positive whole number; found "18" (numeric digits is currently set to 1 digit)." ---rony _______________________________________________ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel _______________________________________________ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel _______________________________________________ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
_______________________________________________ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
