On Mon, 1 Jul 2013 17:23:51 -0400, Farley, Peter x23353 wrote:
>OK, I can see that as a reason. Personally I would prefer an explicit test in
>the code to a remotely-specified out-of-code-body jump, but that's more of a
>style issue than one of substance.
>
>-----Original Message-----
>From: Shmuel Metz (Seymour J.)
>Sent: Monday, July 01, 2013 5:04 PM
>
>>Implementation of standards is only useful if the feature is really
>>needed. I can't think of a reason for needing NOTREADY with classic
>>datasets on z/OS when SYSDSN will tell you whether the file is
>>available or not.
>
>The obvious reason is for detecting end of file without an explicit
>test.
>
True, even though I prefer what Shmuel mentioned elsewhere: an
I/O function that returns a boolean value usable for loop control, as:
do LineNumber = 1 while read( ... )
... avoiding the ugly over-and-under read necessitated by some I/O
functions.
But beyond that, implementation of standards is essential for
portability, both of code and of programmer skill sets. The
argument that "There's a better way of doing that!" whether
idiosyncratic to an implementation or even within the standard
is specious.
And the existence of two stream I/O function packages, one of
which supports legacy data sets but not UNIX files, and another
which supports UNIX files but not legacy data sets is a deplorable
manifestation of Conway's Law.
IBM was once the 800-pound gorilla; what IBM wanted to do
could trump any industry standard. But that was in a different
century.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
