On Fri, 21 Mar 2014 09:21:57 -0700, Skip Robinson wrote:
>    ...
>'NJE with oneself' has no meaning.  ...
>
There's a peculiar tendency of designers and testers to ignore, even
shun, that reflexive boundary condition.

o NJE can't communicate with the local host.
o iconv used to reject input character set and output
  character set the same (it's better now).

TCP/IP, in contrast, is blessedly tolerant.  It even has a
pseudo-hostname, "localhost", for the purpose.

"Why would anyone want to do that?"  Suppose I have a filter
that tailors some JCL and submits it to a remote host.  Suddenly
I want to employ the filter but run the job locally.  Unless the
design was sufficiently modular, I must modify the utility to detect
and treat specially the local host case.

I might want to use iconv to test a file against a character set
for validity by translating it to the same character set.

I might want to test a design without using a remote host.

(One might be prudent to prohibit using a filter where the input
and output files are the same.  That can cause data corruption.
But CMS COPYFILE operating on MDFS files always creates a
temporary file and renames it at the end.  This avoids corruption
in the routine case, and somewhat protects against data loss
even if the system crashes.)

The OP thought he had a valid motivation.

-- gil

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

Reply via email to