🚀 go-pugleaf

In article <41h1jn$7e7@nntp5.u.washington.edu>,
Tim Smith <tzs@u.washington.edu> wrote:
>Eric Smith <esmith@oacis.com> wrote:
>>footprint.  HAH!  Then how can OS/2 run every DOS window as a separate VDM 
>>and every Windows program the same way?  IBM's code is half the size of 
>
>Actually, OS/2 is not all that impressive when it comes to running Windows
>programs in separate VDMs.  Try this: set Word to run in a separate session,
>and then go find the Word icon and double click on it 20 or 30 times.  Watch
>OS/2 crawl, if not totally die.  On NT, it actually works.
>
>Furthermore, if you limit yourself to only 10 or 15 copies of Word (!),
>when I tried this on a 12 MB machine, NT actually managed to get them all
>going without significant swapping.
>
>My speculation is that IBM's separate Windows sessions are a little bit
>too separate.  I think NT is taking advantage of copy-on-write to avoid
>making the sessions completely separate until something actually needs them
>to be completely separate.

   If IBM learned from it's TSS Resident Supervisor experience (1970s),
copy-on-write would apply to OS/2, since it was an ideal way to implement
the UNIX fork/exec paradigm.  This way you can share non-written pages
between lots of "copies" of a process -- such as font info.

>Of course, OS/2 is a lot better than Windows non-NT at running a lot of
>copies of Word in separate sessions.

   Ahh, but the question is:

       How many copies of 'CLOCK' can OS/2 and NT support on a single
display -- and how synchronized are their hands in analog mode.  Since
Windows 3.1, I have only been able to do this by making copies of 'CLOCK'
(like CLOCK1, CLOCK2, etc.).  With only 4 analog clocks, my sped-up 386
has some second hands behind others by about a second.  Even my 1985
Amiga 1000 (with 7.8 MHz 68000) can do better than that.

Al Varney