Hi all,

It was pointed out to me privately that the enhancement I made to the COPY
command (as part of the MPEJXQ1 patch) is inconsistent with DOS and UNIX
systems, which I was aware of.  I modified COPY to support directory names
for the TO= parm.  However, the directory name must end with a slash to tell
COPY that it is a directory.  This differs from UNIX and DOS systems, but is
consistent with COPY's treatment of ".GROUP.acct" as a TO= parm.

Reason:
1  COPY a,b       <-- you know that B is a file
2  COPY a,.b      <-- you know that the resulting file is A.B.logonacct
3  COPY a,./b     <-- you know that ./b is a file
4  COPY a,./dir/  <-- you know the resulting file is ./dir/A

Case 4 is the new scenerio I am discussing.

Cases 2 and 4 have directory or group names and cause COPY to expand the
target name into results you see above.  On a UNIX or DOS system cases 2
and 4 would not need the syntactical hints of a leading "." (followed by
an alpha) to know it is supposed to be a group name, nor the trailing
slash to know it is supposed to be a directory name.  If this was DOS
or UNIX cases 1 and 4 should appear as:

2  COPY a,b      <-- if b is a dir you get ./b/A, if b does not exist you get
                     a file named b.  For B to name a group you would need to
                     use /ACCT/B.
4  COPY a,./dir  <-- if ./dir is a directory you get ./dir/A, if ./dir does
                     not exist you get a file named ./dir as a copy of a

My purpose is to make it clearer when reading scripts or JCL what the
copy command is doing.  This comes at the cost of being inconsistent with
two extremely popular operating systems.

My question to you is do you prefer consistency with DOS and UNIX, or do
you prefer the CI to force upon you a rule that makes the COPY command more
evident?


thanks,
Jeff Vance, CSY

p.s.  I interpret a vote for both ways to mean that you want to be able to
write more readable scripts sometimes but generally prefer the UNIX/DOS
style.

p.p.s. I am not proposing any incompatibility changes, so obviously, whatever
the outcome the CI will continue to support the ".groupname" syntax.

--