Jeff Vance ([log in to unmask]) wrote:
: Gavin wrote:
: >Whoa! Jeff, what's this B.79.05 stuff? Shouldn't HPRELVERSION show
"C.50.xx"?
...
 
 
The major problem, to me, is that I would never trust the reliability
of important software on a value returned (or not returned) from a
CI variable.  Far too many things can go wrong.
 
Heading the list is locally created variables that pre-date HP variables.
For example, a user who (on 4.0) says: let's do a "SETVAR HPRELVERSION
'B.30.45'" ... and ZAP! goes your code that said "if HPRELVERSION
exists, it must be 5.0 or later"!  Or, the site that has AIF:PE
intercepts that *prevent* you from accessing HPRELVERSION (for whatever
reason).
 
No, there is only one reliable method I'm aware of for determining
the MPE XL (and MPE/iX) version:
 
   Loading the build ID directly from system_globals.
 
(Note: this does not require priv mode)
 
 
BTW, the problems with checking for variables and/or commands existence
(or lack thereof) apply to system files as well.  I remember one (unnamed)
vendor's software that checked to see if it was on MPE XL by doing:
 
   if "FREE5.PUB.SYS" is on disk, then we are on a Classic.
 
Buzz! Nope!  On the user's 3000/950 in question, they had created a command
file called FREE5.PUB.SYS which ran DISCFREE.
 
--
Stan Sieler
[log in to unmask]