[Prev][Next][Index][Thread]

Big problems w/ 1.99n5 under Win/95



For some reason my installation of 1.99n5 from the Bleeding Edge over 1.99n
seems to have totally died now under Windows 95 June Test Release.  Before
that though there were problems that mystify me as well.  Using 1.99n I could
run executor if it was the very first thing I did after booting into Windows
95 and could only run it once.  After that trying to run executor again
causes the message "Executor has run out of memory try using a small
-applzone" which seems odd.

I am doing this on an AT&T Globalyst 200 notebook (aka Samsumg Sens 700 w/
486DX2 CPU & 12 MB RAM, WD90C24 video chipset etc) and never had any trouble
at all RUNNING any version of executor from 1.1 on except now I no longer
have DOS.  My Win95 install demanded the primary dos partition and foolishly
I let it have it.  If memory servers this is build 490 of Win95 (typing ver
/r at a DOS prompt gives:  Windows 95 4.00.490 anyway) and I have noticed
that trying to run cwsdpmi on its own says "Protected mode not available" and
then produces a General Protection Fault of sorts.

That at any rate is what it used to do.  Now it is worse.  I either get an
abort with EIP tracebacks - a representative example is below - using either
1.99n or 1.99n5 with applzone 1024 (the only setting which seems not to
immediately provoke the out of memory fail code) 1.99n or 1.99n5; or if I add
-nosplash, an endless stream of the same messages saying "Make sure
$MACDIR/.Rsrc/System is readable and writable and that it has 512 bytes added
to the beginning of it".

Performing Control-Break to cut of this redundant fail code produces this
traceback:

eax=00000000 ebx=00000021 ecx=00000000 edx=00000000 esi=00000039 edi=001ba658
ebp=001ba634 esp=001ba628 cs=00a7 ds=00af es=00b7 fs=0087 gs=00d7 ss=00af 
Call frame traceback EIPs:
  0x00150606
  0x001519da
  0x001546dd
  0x0014d4da
  0x00149f6c
  0x0008b815
  0x000a705e
  0x0008b845
  0x000a705e
  0x0008b845
  0x000a705e

Another example (w/out the -nosplash) is:

eax=00000c00 ebx=00000c00 ecx=000c0000 edx=00000000 esi=00265636
edi=00000000 ebp=001c21c8 esp=001c217c edi=00000000 ds=00af es=00b7
fs=00d7 gs=00d7 ss=00b7
Call frame traceback EIPs:
0x00099cee
0x0006f28a
0x00074af2
0x00070d04
0x000adbad
0x000ae63e
0x0099fcc4
0x0014b168

Division by zero at EIP=00099f6 shows up as well.  I am at a loss over all
this since 1.99m worked with the February BETA and this configuration worked
for a short while as well it seemed.  Does anyone have any suggestions how to
finesse the right settings on a DOS program objects properties/memory tab?
 Anyone else using build 490 of Windows 95 who is having trouble with
Executor or anything else or conversely are there any wonderful success
stories to report?

Also, not a bug but an incompatibility to possibly look at if there is time
in the HACKATHON, I have noticed that PGP 2.3a aborts as well, shortly after
its first splash screen with credits for Phil Zimmerman et al. and I can only
assume it has something to do with PGPs own emulation of floating point
processor code for RSA algorithm calcs.  Sorry I cannot furnish an EIP
traceback for this one.  It has been some time since I actually tried to run
it and since that was under 1.99m prior to my joining this list or even
knowing about it I just wiped the HFV I had it installed in.

And finally, much has been made of full System 7 support "some day"
presumably post 2.0 wherein we could feasibly buy the Mac/OS from Apple and
install it over Executor.  Well is anything like this presently possible with
System 6 from Apple?  Or am I mistaken to believe versions prior to 7 were
sold shrinkwrapped rather than pre-built onto the systems themselves?  I seem
to recall having read somewhere once that System 7 was in fact the first you
could just walk in a store and buy all wrapped up in a box.

All in all, great work ARDI.  If only Win 95 were as good a BETA right now
;-)

John Clark - who considers going back to DOS 3.3 a viable alternative!!


Follow-Ups: