RE: Should this pipe run
Glenn, I hate to disagree with you. But.
The commit mechanism has a lot more to it than Glenn lets on.
For < no file|, the nonextant file is caught in the syntax check
routine; not a single stage has started (and nor will any). Commits
do not enter into this. True, had I done commit first, it is unlikely
I would have bothered with a separate syntax check facility.
It is OK for a stage to exit without committing even when it starts at
a negative level. It is the non-zero return code that triggers the
pipeline being abandoned.
>sfs with the SAFE option commits to level 1 before committing the
unit of work; it rolls back in the event of failure. It is a built-in
program.
The example for PIPE AHELP COMMIT shows how a REXX program can check
the return code from other programs as to whether it should continue,
even if it does not want to perform any checks of its own before
committing to level -1.
By now, I'm sure Bob would be a lot less confused if he had just read
the paper.
j. [EMAIL PROTECTED]
