Scott asks:
> > > System abort 1144 from subsy 101
> > :debug
> > = errmsg (#1144, #98)
>
> Well SA 1144 is indeed a SA for the reason you noted, but I am
> curious about your = errmsg statement... memory management is
> actually subsys 101, not subsys #98... typo?
Um, you didn't take my internals class, did you?! :)
All system failures are *supposed* to be in set 98 (decimal).
Why that was decided, I don't know. Perhaps to facilitate
producing a list of all possible system failure messages (i.e.,
dump set 98 from the system catalog)
It "Sure Would Be Nice" if both of the following were true:
1) the system failure message was in the specified
set (#101 in this case)
i.e., so =errmsg (#1144, #101) would *WORK*
(so we don't have to teach people about the magic number 98)
2) all system failures would be gathered together into
set 98 as well (since most are there already)
(many aren't in the catalog anywhere!)
I try to submit bug reports for whenever I find a system
failure that isn't in the catalog (in set #98), but I often
forget.
ok, while we're at it:
3) if *EVERY* subsystem had it's name available via:
=errmsg (#32765, <subsystemnumber>)
Many are there...but not all.
No, *I* didn't choose 32765...I'd have chosen the incredibly
obvious value of 0...because an hpe_status with an info of
0 and a subsystem of non-0 is *illegal*, and would be a
convenient method of saying "hi, I'm subsystem number X".
(It's illegal because we wanted a single, simple way of
saying "ok" ... and that's 32-bits of 0.)
I tell my students that there are a small number of magic
values to memorize:
$c0000000 (system globals address)
#98 (= errmsg (<systemfailure#>, #98)
2 (= errmsg (<CIERR number>, 2)
8 (= errmsg (<FileSYstemError number>, 8)
#32765 (=errmsg (#32655, subsystemnumber)
0, 1, 2 (values of ccG, ccL, ccE ... you're GLEeful when you remember them)
--
Stan Sieler [log in to unmask]
http://www.allegro.com/sieler/
|
