Conference evms::y2k

    
    Here are a few quick and dirty techniques I ran across during the
    AMACRO checkout.  These may seem like a bit of pedantry, but a daunting
    task like line-by-line checks of all this code can cloud the mind a bit.
    Your mileage, of course, may vary.
    
    Generally speaking, this will go faster if you'll do a bit of up front
    work.  Spare yourself the agony of checking each listing or source file
    fresh off a result disk or out of the master pack.  What may work for
    one facility may not work for another.
    
    1) Use source files whenever possible.  If you're using listing files
    for some reason, copy them off a result disk and use an editor to
    eliminate things you don't need to look at, like generated machine code
    listings and lines full of library or header file references (C
    #includes, BLISS LIBRARY and REQUIRES, etc.)
    
    2)  Look for opportunites to make the VMS SEARCH command do a little
    work for you.  For example, a module might contain a lot of function
    calls, subroutine calls, or assembly language code you already know to
    be Y2K safe.  These modules can be pared down with judicious use of the
    SEARCH command using a /MATCH=NOR qualifier.  This same technique can
    spare you looking at calls to system services that have already been
    judged bulletproof, like $GETTIM.
    
    3) Skim over comments to make sure they don't contain any useful hints,
    then throw them out before doing your line-by-line check.
    
    DFW

    I found a gotcha with searching for uses of $FAO and !%D, !%T.
    
    When an expected occurrence of !%D didn't show up in the search output
    for SDA, I took a closer look at the $FAO call where I knew it was
    used.
    
    It didn't show up because I had used !17%D to only display date, hours
    and minutes but not seconds or hundredths (because object fils and STB
    files only contain their creation date/time to that granularity).
    
    So !%D would never find them. And you don't want the search tool to
    blindly search for %D and %T, since those will find %d in C printf
    calls, and %DECLARED and %THEN in Bliss source.
    
    But you *can* do searches that will limit the noise generated. For
    example:
    
    	$ SEARCH *.* "%D,"%DECLARED"/match=xor/exact
    
    will find %D but not %d or %DECLARED
    
    - Richard.

�     It didn't show up because I had used !17%D to only display date, hours
�     and minutes but not seconds or hundredths (because object fils and STB
�     files only contain their creation date/time to that granularity).

Try using the "FIND" and/or "FIND/REPLACE" program (which I got from the
enet somewhere - the toolshed library???). This program allows the use
of more complex search regular expressions - you could find this type of
$FAO string with:
	$ FILE input_file "!\d\.\%\[Dd]"
which means:
- ! => find a literal "!"
- \d\. => find zero or more decimal digits
- \% => find a literal "%" (the % character by itself means match any
        single character
- \[Dd] => find an upper or lower case D

You xcould also use
	$ FIND input_file "!\(\d\.\)\%\[Dd]
where the extra "\(" and "\)" delimit a 'group' which can be referred to
later so that (for example in a FIND/REPLACE command) the number can be
processed in some way.

Any help?

Cheers,

Don

    re .7
    
    What about
    
    $ SEAR/MAT=AND file "!","%D"
    
    Perhaps these problems will be the trigger that finally encourages
    someone to upgrade SEARCH to support pattern matching.