I have a problem that I can solve with REXX
but
would like ideas on how DFSORT, might be able to
perform the same function, since I am talking around 1,000,000 plus records.
As input I have a dataset with a list of dataset
names from a catalog (really several catalogs),
in columns 1 thru 44. What I want to do is
include or exclude dataset names in the file
based on patterns like you might enter on ISPF 3.4
xxx.yyy.*.zzz.**
ttttt.*.rrrrrrr.**
To make life easier a qualifier will be a value,
* or **. It would be great if the process could
accept bb*, but not a requirement. Note that
there could be 100+ patterns to match, not just a few that could be hardcoded.
I know how to parse the DSN into qualifiers
painfully (is there a better way?)
PARSE=(%01=(ABSPOS=01,ENDBEFR=C'.',FIXLEN=8),
%02=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%03=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%04=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%05=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%06=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%07=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%08=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%09=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%10=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8),
%11=(ENDBEFR=C'.',ENDBEFR=C' ',FIXLEN=8)),
But how to get sort to accept the * or ** to
match anything? I could maybe see a pre pass to
convert the trailing ** into a series of
*.*.*
But I then there is the
zzzz.**.pppp.vvvv.** that would just ruin my
day. Lets say that form is not valid.
Any ideas? Code would be great, a concept or
process that DFSORT can perform is what I am looking for.
Ken Leidner [email protected]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
