koren@hpfelg.HP.COM (Steve Koren) (07/11/90)
Here is an update on SKsh 1.5 and a question at the end for folks
who know something about the 3000/AmigaDos 2.0.
SKsh 1.5 update:
---------------
I thought I owed SKsh users a status update on 1.5 since it has been
delayed over a month now. The delay was caused by two problems; first,
a bug in Lattice 5.05 that it took me a while to find a fix for, and
second, my Amiga upped and died. Both problems are more-or-less
resolved now, and 1.5 is back underway. I hope to have it out in
less than two weeks from Mon 9 Jul 90. It provides more enhancements
than any other upgrade to date. Sorry for the delay!
SKsh and Amiga 3000's
---------------------
I received a phone call yesterday from a fellow who was running
SKsh 1.4 on his A3000. He said a few interesting things. First,
that if you have 1 meg of chip memory configured on the 3000, that
Sksh works fine under either AmigaDos 1.3 or AmigaDos 2.0. And
second, that SKsh crashes pretty seriously if there is 2 megabytes
of chip ram configured.
This brings up a question for 3000 gurus. SKsh 1.4 does about
98% of its memory allocation using the Lattice malloc() call, and
the other 2% using AllocMem(blah,MEMF_CHIP) to get chip memory
(for a system call that wants it to be there). Should any of this
care about extra chip memory? Not as far as I understand: the Lattice
malloc() just returns whatever it can find, and AllocMem() ought to
be smart enough to deal with extra chip memory.
So what in the heck could be the problem? Anyone have any ideas about
why something should act differently in those cases? I know that
SKsh does fine with either 512K or 1 meg of chip ram (it uses so
little chip ram, only a few bytes, that it doesn't matter).
Help!?! I alas don't have a 3000 to test with, so I can't do much
in the way of finding this problem, but I'd like to resolve it
before 3000s become widely used. Anyone seen a problem like this
before? I'm clueless at the moment.
- steve