On Tue, 20 Mar 2012 10:54:50 +0100, Thomas Berg wrote:
>The only alternative (as seen from the principle of least astonishment) I can
>think of is using the explicit option "NULL" - assuming that value is never a
>"real" option.
>
In some contexts, such an option is unjustified. Designs should
attempt not to restrict value spaces.
>BTW, In rexx if you check for an existing parm would "rexxfunc(arg1,,arg3)"
>have a *non-existing* arg/parm nr 2 but "rexxfunc(arg1,'',arg3)" would have
>and *existing*, but "empty" arg/parm nr 2.
>
Empirically, yes:
5 *-* trace R
7 *-* Junk = rexxfunc(arg1,,arg3)
13 *-* rexxfunc:
14 *-* return( arg( 2, 'Exists' ) )
>>> "0"
>>> "0"
8 *-* Junk = rexxfunc(arg1,'',arg3)
13 *-* rexxfunc:
14 *-* return( arg( 2, 'Exists' ) )
>>> "1"
>>> "1"
9 *-* junk = arg( 1, )
>>> ""
10 *-* junk = arg( 1, '' )
10 +++ junk = arg( 1, '' )
IRX0040I Error running ./fooargs, line 10: Incorrect call to routine
Since you ask the question, I checked the doc at:
http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/ikj4a3a0/4.3.4
I can see that it doesn't make this explicitly clear in the examples
given. Is an RCF merited?
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
