LLACROIX@carleton.edu.UUCP (01/27/87)
I believe that VMS system services avoid checking the # of arguments for speed
considerations. What you really want to do for argument checks in the kernel
are to avoid trashing the system or making a breech of security. Anything else
is overhead that applies to all users on all systems, large to small. Not
really desirable.
Besides, someday everyone will have access to smart editors and program
verifiers so that this kind of error will be avoided from the start (or at
least easily discovered)...VMS might still be around by then, too. Maybe.
-----
Les LaCroix | csnet: llacroix@carleton.edu
SPSS, Inc. | uucp: ...{ihnp4,decvax}!stolaf!ccnfld!llacroix
402 Washington Street |
Northfield, MN 55057-2027 | [standard disclaimers apply]sasaki@HARVARD.HARVARD.EDU.UUCP (01/30/87)
The reason that system services have a fixed length is entirely one of programming style. The folks who wrote VMS were used to RSX-11M, which had system services with a fixed number of arguments. RSX-11D (and IAS too) had system services with variable argument lists. Speed is not a major consideration, it doesn't take that much time to figure out how many arguments there are. As an example, take the new logical name services. Logical names get translated all of the time, yet both crelnm and trnlnm have item lists which have to be figured out every time. ---------------- Marty Sasaki uucp: harvard!sasaki Ziff Davis Technical Information Co. arpa: sasaki@harvard.harvard.edu 80 Blanchard Road bitnet: sasaki@harvunxh Burlington, MA 01803 phone: 617-273-5500