Talk:Permutations with repetitions

REXX implementation limits

Most REXX interpreters have implementation limits on the length of a REXX clause.

For Regina REXX, it depends on the release level.

For Regina REXX 3.3,   3.4,   3.5, REXX gives a SYNTAX error:

Error 12 running "C:\xxxxxx.REX", line 29: [Clause > 1024 characters]

For Regina REXX 3.6,   3.7, REXX "crashes" somewhere less than a statement length of 1932,
I haven't pursued this to find the exact byte count.

{By crashing, I mean Microsoft Windows displays a small error "window":

Regina Rexx Launcher (x86) has encountered a problem

and needs to close.   We are sorry for the inconvenience.

(more boilerplate follows).

For what's it worth, I believe REXX should raise the SYNTAX condition.

[The reason I mention the older versions of Regina REXX is that Regina REXX 3.3 uses less
virtual memory for those REXX programs which use a lot of stemmed variables (possibly a
memory leak), so the use of the older Regina release is forced in lieu of a fix.

For PC/REXX and Personal REXX, it depends on what the RXISA (SET environmental variable).
These two REXXes either give:

Error 11 on line 29 of c:\xxxxxx.REX" Control stack full

Memory full: symbol

Error 5 on line 29 of C:\xxxxxx.REX: Machine resources exhausted

somewhere between 568 and 597 bytes.

It should be noted that most people use the maximum for the RXISA   (which is 40K).

As far as I can tell, R4 and ROO have no (real) limitation for the length of an INTERPRET instruction,
and that is probably true of CMS REXX and TSO REXX   (it's been too long since I used those
two flavors of REXX).

-- Gerard Schildberger (talk) 20:52, 12 May 2013 (UTC)

---