[comp.windows.ms] Windows 3.0/Multi-tasking/286 & DOS window shells?

defaria@hpclapd.HP.COM (Andy DeFaria) (12/15/90)

I have  Windows  3.0 installed on  my system.   I've been told that it does
multi-processing.  I have  a 286 chip though.

Anyway,  when I'm  in  Windows I don't see   any  multi-processing at  all.
Whenever a Window application is  running I  get the little hourglass mouse
cursor and I can't continue to work in another window.

If I  can get  multi-processing in Windows 3.0 going  on  a 286 chip then I
would be a lot  happier but there is  still a big hole  that I see  in this
environment:

On my workstation (Unix/X)  I  have some window smart  applications running
(x11notes, datebook, etc).  Any software that I want to  run that is window
dumb I do in a hpterm.  Now, translating that  into the DOS  world, I would
like to be able  to run a DOS shell  in a window on  my PC.  I don't expect
that I could run, say,  Turbo Pascal from  the DOS window  shell but  if  I
could  have multiple DOS window  shells   running  simple  DOS commands and
programs that don't assume  that the monitor is theirs  to destroy, I would
be alot more productive.  Is  it possible to run DOS  in a Microsoft window
using Windows 3.0/286  chip?  I see the configuration  options in  the  PIF
editor to run things like DOS shells in a window but it says that it is 386
only options and my 286 ignores them.

I had Windows 2.0 for a  time and I  was able to  run COMMAND.COM  from the
MSDOS "Run" selection.  It put up a  window that I  could type DOS commands
in.  How is this done in 3.0?

If it can't be done in 3.0 then isn't this a loss of functionality?

bcw@rti.rti.org (Bruce Wright) (12/20/90)

In article <27220002@hpclapd.HP.COM>, defaria@hpclapd.HP.COM (Andy DeFaria) writes:
> I have  Windows  3.0 installed on  my system.   I've been told that it does
> multi-processing.  I have  a 286 chip though.
> 
> Anyway,  when I'm  in  Windows I don't see   any  multi-processing at  all.
> Whenever a Window application is  running I  get the little hourglass mouse
> cursor and I can't continue to work in another window.

This happens when the application is somewhat mis-behaved and loads
the hourglass cursor while it does some compute-bound operation, but
then doesn't relinquish the CPU to other apps in the system until it's
through.  Windows _is_ multi-tasking, but only for cooperating
processes;  a single ill-behaved task can cause the behavior you
describe.  In general, if the application is going to have the
hourglass displayed for more than a very few seconds (say 5), it
_should_ relinquish the CPU and run in the background, only displaying
the hourglass while it's the window with the input focus.  Too many
applications _don't_ behave very well though.

> If I  can get  multi-processing in Windows 3.0 going  on  a 286 chip then I
> would be a lot  happier but there is  still a big hole  that I see  in this
> environment: [...]
> 
> I had Windows 2.0 for a  time and I  was able to  run COMMAND.COM  from the
> MSDOS "Run" selection.  It put up a  window that I  could type DOS commands
> in.  How is this done in 3.0?

It can't be done on a 286 in 3.0.  You can only swap to full-screen
DOS mode, not the DOS window like you had in 2.11.  

> If it can't be done in 3.0 then isn't this a loss of functionality?

Yes, of course it is.  The consensus seems to be that this isn't an
unreasonable trade-off, given that the new method for running DOS
programs in a window allows all sorts of apps to run that used to
write all over the screen and confuse Windows (or at least the user
running Windows).  In addition, the new approach allows the DOS
app to get a full 640K of memory, rather than just whatever was
available after Windows and its apps were loaded.

Unfortunately it can't work on a 286 :-(.  Worse, there would be
problems running _any_ DOS window in protected mode on a 286 even 
if the DOS app were running in the lower 640K of memory and _didn't_
do anything like direct screen writes.  It would be possible to run
something similar to a Windows 2.11 DOS window by putting the system 
in real mode while the DOS app was running, but performance would be 
terrible.  Under Windows 2.11, Windows was running in real mode anyway,
so there wasn't much of a penalty for having the DOS window.

I'm afraid that you'll either have to live with it or upgrade the
CPU to a 386 of some flavor - a 386sx shouldn't be too different in
cost from a 286 at this point.

defaria@hpclapd.HP.COM (Andy DeFaria) (12/21/90)

>/ hpclapd:comp.windows.ms / bcw@rti.rti.org (Bruce Wright) / 12:08 pm  Dec 19, 1990 /

>This happens when the application is somewhat mis-behaved and loads
>the hourglass cursor while it does some compute-bound operation, but
>then doesn't relinquish the CPU to other apps in the system until it's
>through.  Windows _is_ multi-tasking, but only for cooperating
>processes;  a single ill-behaved task can cause the behavior you
>describe.  In general, if the application is going to have the
>hourglass displayed for more than a very few seconds (say 5), it
>_should_ relinquish the CPU and run in the background, only displaying
>the hourglass while it's the window with the input focus.  Too many
>applications _don't_ behave very well though.

You  got that right!    Considering the   application  in  question  are MS
supplied with Windows 3.0, are you  saying that even MS applications aren't
well behaved?

>I'm afraid that you'll either have to live with it or upgrade the
>CPU to a 386 of some flavor - a 386sx shouldn't be too different in
>cost from a 286 at this point.

My 286 is a loner from work so it cost me $0.00.  Do you know  where  I can
get a 386sx for the same?!?

bcw@rti.rti.org (Bruce Wright) (12/22/90)

In article <27220003@hpclapd.HP.COM>, defaria@hpclapd.HP.COM (Andy DeFaria) writes:
> >/ hpclapd:comp.windows.ms / bcw@rti.rti.org (Bruce Wright) / 12:08 pm  Dec 19, 1990 /
> 
> >This happens when the application is somewhat mis-behaved and loads
> >the hourglass cursor while it does some compute-bound operation, but
> >then doesn't relinquish the CPU to other apps in the system until it's
> >through.
> 
> You  got that right!    Considering the   application  in  question  are MS
> supplied with Windows 3.0, are you  saying that even MS applications aren't
> well behaved?

There's been no secret for some years now that Microsoft applications
are often not as well-behaved as they ought to be.  This is just as
true in Windows as in vanilla DOS.

In other words, this is stale news.

However I might point out that it is possible that your system is
underconfigured as far as memory goes - if a large application has to
reload segments from disk very often then it can take much longer to
perform some operations than the developer expected, and he/she might
not have provided for allowing a task switch in those places that
'ought' to be fast.  But in limited-memory situations, you can easily
get situations where _nothing_ is fast because Windows is thrashing
the disk.  It can also be helpful to run with a disk cache.

Moral:  Windows likes memory.

> >I'm afraid that you'll either have to live with it or upgrade the
> >CPU to a 386 of some flavor - a 386sx shouldn't be too different in
> >cost from a 286 at this point.
> 
> My 286 is a loner from work so it cost me $0.00.  Do you know  where  I can
> get a 386sx for the same?!?

I'd submit that the 286 you're using is therefore not 'your' 286.  If
you're using it and Windows 3.0 for company purposes they may agree
that it would be worthwhile to swap it out for a 386sx or get one of
the 386 plug-in cards for the existing machine (usually a costlier
option...), assuming anyway that those company purposes are more
demanding than using Terminal to read E-mail ;-).

I can certainly agree that it's annoying to lose a feature;  this
was one that I regretted losing as well.  I have an old V20-based
portable PC with a 10MB external hard drive that I sometimes use 
for Windows programming if I'm away from home (stop laughing out 
there :-) and it was really nice to be able to fire up small, well-
behaved DOS apps on that machine.  The performance is even not as
bad as one might think;  the only problem is compiling, but if I'm
at the beach I can do some programming early in the morning and 
after lunch and start up the compiler while I swim or walk on the
beach :-).  Not a very productive work machine, perhaps, but still
one that you can use to hack on 'fun' programs with (sort of a
bus driver's holiday :-).

						Bruce C. Wright