On 20 Feb 2013 13:37:40 -0800, in bit.listserv.ibm-main Leslie wrote: >On Wednesday 20 February 2013 02:15:51 Thomas Berg wrote: >> > It's not the features that are bad in those instances, but rather the >> > syntax for requesting the features; that syntax is about as far from >> > the purported English-like character of COBOL as you can get. >> > >> > >I can't immediately see that (except maybe COMPUTATIONAL-n). > In COBOL 2000 the Computational usages have been supplemented by
None of the BINARY usages below are recognized by the Enterpise COBOL compiler even though they are in the 2002 standard and some would be useful for dealing with SMF records. The BINARY usages, especially the new floating point usages that would be very useful in dealing with other languages that recognize IEEE floating point, could simplify inter-language calls with C and Java. It is ludicrous that COBOL, the business language does not recognize Decimal Floating Point and the new rounding facilities even though Mike Cowlishaw of IBM Rexx renown was one of the major pushers for the facility and it is in both C/C++ and PL1 as well as Java. Clark Morris >BINARY * >BINARY-C-LONG SIGNED >BINARY-C-LONG UNSIGNED >BINARY-C-LONG or >BINARY-CHAR SIGNED >BINARY-CHAR UNSIGNED >BINARY-CHAR or >BINARY-DOUBLE SIGNED >BINARY-DOUBLE UNSIGNED >BINARY-DOUBLE or >BINARY-LONG SIGNED >BINARY-LONG UNSIGNED >BINARY-LONG or >BINARY-SHORT SIGNED >BINARY-SHORT UNSIGNED >BINARY-SHORT or >COMP * >COMP-1 * >COMP-2 * >COMP-3 * >COMP-4 * >COMP-5 * >COMPUTATIONAL * >COMPUTATIONAL-1 * >COMPUTATIONAL-2 * >COMPUTATIONAL-3 * >COMPUTATIONAL-4 * >COMPUTATIONAL-5 * >COMPUTATIONAL-X >DISPLAY * >DISPLAY-1 * >FUNCTION-POINTER * >INDEX * >NATIONAL * >OBJECT REFERENCE class-name-1 * >PACKED-DECIMAL * >POINTER * >PROCEDURE-POINTER * >PROGRAM-POINTER >SIGNED-INT >SIGNED-LONG >SIGNED-SHORT >UNSIGNED-INT >UNSIGNED-LONG >UNSIGNED-SHORT >so this should no longer be an issue (unless they are prohibited by hide-bound >installation standards). (* indicates those recognized by z/COBOL.) >> > >> > If you're just learning COBOL, the magic numbers 77 and 88 totally >> > obscure the intent; I consider them to be worse than COMPUTATIONAL-n in >> > that regard. >> >> Do you in this regard prefer, e g, that: >> >> 01 NAME1 PIC X. >> 88 ONE VALUE '1'. >> 88 ZERO VALUE '0'. >> >> - instead be: >> >> 01 NAME1 PIC X. >> WHEN VALUE '1' SETTRUE ONE. >> WHEN VALUE '0' SETTRUE ZERO. >> >> ? >> >> But I can't see level number 77 be much confusing, out of line of "normal" >> COBOL and maybe superfluous but not much other than that. > 66, 77 and 88 were perhaps an unfortunate choice by the Codasyl > Committee, >but simple code inspection makes their purpose clear even to a neophyte after >a few moments. In COBOL 2000 we now have level 78 as well, to identify >constant identifiers: >78 identifier-1 VALUE IS literal-1 . >an alternative that may be preferred (I do) is use of the CONSTANT keyword: >01 identifier-2 CONSTANT [ IS GLOBAL ] AS [literal-2 | LENGTH OF >identifier-3 | BYTE-LENGTH OF identifier-4]. > All issues with level numbers and usage clauses may be quickly resolved > by >looking at the COBOL Language Reference manual (unless one has an aversion to >reading it). > >Leslie > >---------------------------------------------------------------------- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
