Fw: COBOL II and COMPS
All of this is "fixed" in the 2002 COBOL Standard where *NEW* binary data-types are introduces (with NO picture clause - just something indicating "size" and whether or not they are signed). An existing SHARE requirement exists for IBM to implement these. (It even includes a one-byte binary field - which has always required "odd" coding in COBOL). If your site wants to create a (marketing) REQUEST to indicate a need/desire for these, contact me off-list at: wmklein ix.netcom.com and I can give you a SHARE requirement number to add to your "MR" request. "john gilmore" <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>... > Bill Klein writes: > > > > >On the other hand, the only time that a COMP-5 data item works differently > >from a BINARY or COMP (or COMP-4) item (on the mainframe) is *IF* you try > >and use values greater than the PICTURE clause allows. This is an "ugly" > >thing to do and is NEVER recommended. > > > > My response is that I strongly recommend that just this be done, and here we > have what is not so much an intellectual as a visceral, attitudinal > difference. > > In PL/I, fror example, one writes > > declare binary_counter binary fixed(31,0) ; /* fullword counter */ > > declare packed_counter decimal fixed(7,0) ; /* seven-digit packed counter */ > > In the first declaration (31,0) denotes the number of binary digits and the > number of binary digits following the binary point; and in the second > declaration (7,0) denotes the nu,mber of decimal digits and the number of > decimal digits following the decimal point. > > COBOL does not make this distinction. Picture specifications are decimal > and decimal only EVEN FOR BINARY DATA TYPES. Unfortunately perhaps but > unsurprisingly, the powers of 2 and those of 10 are different; and a decimal > picture specification cann therefore newver accurately represent the > precision of a binary data type: It must significantly understate or, much > worse, overstate, it. > > COBOL is the only statement-level procedural language that has come to terms > only very grudgingly and still oinly very partially with binary data types > that the hardware, be it an Intel microprocessor or an IBM maniframe, makes > available. > > -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: COBOL II and COMPS
Bill Klein writes: On the other hand, the only time that a COMP-5 data item works differently from a BINARY or COMP (or COMP-4) item (on the mainframe) is *IF* you try and use values greater than the PICTURE clause allows. This is an "ugly" thing to do and is NEVER recommended. My response is that I strongly recommend that just this be done, and here we have what is not so much an intellectual as a visceral, attitudinal difference. In PL/I, fror example, one writes declare binary_counter binary fixed(31,0) ; /* fullword counter */ declare packed_counter decimal fixed(7,0) ; /* seven-digit packed counter */ In the first declaration (31,0) denotes the number of binary digits and the number of binary digits following the binary point; and in the second declaration (7,0) denotes the nu,mber of decimal digits and the number of decimal digits following the decimal point. COBOL does not make this distinction. Picture specifications are decimal and decimal only EVEN FOR BINARY DATA TYPES. Unfortunately perhaps but unsurprisingly, the powers of 2 and those of 10 are different; and a decimal picture specification cann therefore newver accurately represent the precision of a binary data type: It must significantly understate or, much worse, overstate, it. In the tables |---| | hardware capacity | |---| | b | unsigned | signed | |---| | 1 | 0 <= Y <= 255 |-128 <= y <= +127 | | 2 |0 <= H <= 65,535 | -32,768 <= h <= +32,767 | | 4 | 0 <= F <= 4,294,967,295 | -2,147,483,648 <= f < = +2,147,483,647 | |-|-| |--| |COBOL picture capacity | |--| | b | unsigned | signed | |--| | 1 | 0 <= Y <= 99 | -99 <= y <= +99 | | 2 | 0 <= H <= 9,999 | -9,999 <= h <= +9,999 | | 4 |0 <= F <= 999,999,999 | -999,999,999 <= f < = +999,999,999 | |--| COBOL is the only statement-level procedural language that has come to terms only very grudgingly and still oinly very partially with binary data types that the hardware, be it an Intel microprocessor or an IBM maniframe, makes available. John Gilmore Ashland, MA 01721 U.S.A. _ Dont just search. Find. Check out the new MSN Search! http://search.msn.click-url.com/go/onm00200636ave/direct/01/ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
