If the SHARE requirements for implementing the new BINARY, FLOATING-POINT and BIT usages in the 2002 COBOL Standard were implemented where the new floating point usages would be in IEEE floating point more efficient code could be generated. The new binary usages (BINARY-CHARACTER, BINARY-SHORT and BINARY-LONG) explicitly are for half-word and word sized fields and thus avoid the TRUNC problems. These usages plus USAGE BIT would allow handling of ALL SMF records without going through contortions. There may also now be a usage for double word. The new floating point usages could eliminate the conversions when interacting with JAVA and C programs thus improving web performance. There are business programs that use files with bit switches. I have worked with a system that had them on open account, product and customer records. It was an in house system. There are probably other additions in the 2002 COBOL standard that would allow the writing of code that could be generated more efficiently with greater ease and clarity for the writer.
In addition, the date conversion routines are inefficient and I have written COBOL code that showed a noticeable difference in CPU time for the conversions (Gregorian to and from Julian or either to and from COBOL day number). This was in conjunction with a Year 2000 project. The generated code did not to seem to be that much greater than the generated CALLS when using the COBOL date FUNCTIONS to get day number. ---------------------------------------------------------------------- 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
