On Fri, 27 Nov 2009 16:39:32 +0100, Miklos Szigetvari wrote:
>
>The input is already in a HFS mounted drive, the simplest would be to
>use it directly, I have tested it ,
>and with //SYSUT1 DD PATH=.. didn't worked so I changed  a while ago to
>copy the HFS to
>a temporary MVS dataset and unterse.
>
I believe that in the earliest days of OpenEdition PATH=
was supported by allocation (necessary I believe, so Binder
could read SYSLIN and generate program objects in SYSLMOD
in Unix files and directories).  But there was no access
method support for PATH=; the utilities needed to detect this
and use BPX1* calls to perform Unix file I/O.  So many utilities
checked and prudently prohibited use of PATH= for SUSUT1, SYSUT2,
etc.

Nowadays, there is excellent access method support for Unix
files and (more recently) directories.  But many utilities
have not relaxed the now needless check (IEBGENER is a
welcome exception).  For example, AMATERSE fails on Unix
files for its compressed I/O, a restriction reasonable only
for the uncompressed which is manifestly device dependent;
and Rexx likewise fails with a Unix directory as SYSEXEC.

In either case, the check can be circumvented by precatenating
an empty Classic PS or PO data set.  This shows that the check:

o Is unnecessary because things work fine if it's circumvented.

o Fails its intent because it neglects to check additional
  catenands.

<RANT> What became of the concept of device-independent I/O?
The utility should simply OPEN the DCB supplied by the programmer
and check for errors returned by the access method, not attempt
to outsmart the access method.

It has been written here that there are performance benefits
available to device-savvy utilities.  I don't believe that's
what motivates AMATERSE and Rexx.  And, regardless, in those
cases where the utility encounters unanticipated characteristics,
it should still fall back to attempting access method I/O </RANT>

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to