On Tue, 2 Feb 2021 14:57:28 +0000, Seymour J Metz wrote: >The confusing part is is not the way compound variables work, but the way >value() works. > The doc must be read carefully, but then is it confusing? Both TSO/E Rexx and Regina say some paraphrase of" "The argument of VALUE() is a string which is the name of a symbol". I take this to imply these are equivalent barring syntax error: interpret 'say' X and say value( 'X' ) (If I'm wrong, then it *is* confusing.)
>In OOREXX there is syntax for a tail that includes lower case characters, e.g., > > foo=.stem~new > foo['bbb']=7 > Does OOREXX require preallocation of compounds with a "new" method? >________________________________________ >From: Charles Mills >Sent: Monday, February 1, 2021 7:09 PM > ... >I pictured it kind of like C or COBOL multi-dimensional arrays. I pictured >Rexx A.B.C.D being essentially analogous to C language A[B,C,D] or COBOL >A(B, C, D) albeit with "associative subscripts." > >But it really is more like a one-dimensional array than an n-dimensional >array. .. > FORTRAN doc when I learned FORTRAN specified the algebraic relation of multiple subscripts and single subscripts, and that single subscripts might exceed the declared bounds provided that the algebraic combination lay within the storage that array occupied. Matrix algebra routines exploited that fact. Awk treats A[ i, j, k ] as A[ i SUBSEP j SUBSEP k ] where the programmer can control the value of SUBSEP. >A. is special; it is "all the possible tails of A" but A.B. is just "A.B >plus a period." > I believe that facility is peculiar to Rexx, and would make it difficult to provide an enumerator, common in other languages with associative arrays. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN