On Sat, 30 Nov 2013 16:55:58 -0600, Anthony Babonas wrote:

>Don't forget the hyphen and x'C0'.
> 
Hyphen is strange.  JCL allows hyphen in data set names in some
contexts; reports it as a syntax error elsewhere.  I believe this is
documented.

ISPF LM services allows hyphen in member names in some contexts;
reports it as a syntax error elsewhere.

Is there any rationale for this erratic behavior?  Conway's Law?

I understand that x'C0' was a mistake.  A coder wrote some
sequence of CVD and UNPK, forgetting to repair the sign
nybble (perhaps in SVC numbers?)  It was immediately
recognized as too deeply embedded to repair, and institutionalized.
I'm not an assembler programmer.  Mentally, I can't envision
the exact instruction sequence or should have been.

With a brief exposure to MVS, I started to learn CMS.  I was
shocked (briefly) to learn that file names might begin with
numeric digits; in fact be entirely numeric.  Why not in
OS/360 data set names?  In an era of severe storage and CPU
cycle constraints, the lexical analyzer would have been simpler
for not needing to treat the first character specially.  Would
allowing numeric data set names have introduced a syntactic
ambiguity in JCL or elsewhere?  Member names couldn't
unambiguously be numeric because of GDG levels.

Earlier in this thread, someone did some arithmetic showing
that the current data set name conventions allow more data
sets than could be stored on any current or reasonably
envisioned volume.  Therefore there's no rationale for
enlarging the name space.  Beware such arguments based on
name space cardinality -- I suspect that there would be
enough available data set names if only letters in the first
third of the alphabet were permitted, so why allow any more?
Becayse some programmers like the flexibility.  I would like
the flexibility of lower case alphabetics.

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to