home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Magnum One
/
Magnum One (Mid-American Digital) (Disc Manufacturing).iso
/
d21
/
dvsi1_21.arc
/
WHATSNEW.DOC
< prev
next >
Wrap
Text File
|
1991-10-09
|
10KB
|
246 lines
A list of the changes....
------------------------------------------------------------------------
October 9, 1991: Version 1.21 released
Version 1.20 didn't stay around too long. I found a few annoying
bugs after releasing it, so I figured I'd quickly get a maintenance
release out before anyone downloaded it. I had intended on fixing
these bugs for 1.20, but didn't get around to them.
BUG FIXES:
PERF:
If Perf was loaded from a startup script, it was possible
for a "Divide by 0" error to occur.
PS, DF, PERF, MEMSTAT:
Sometimes the information would scroll outside the viewport.
This should no longer happen.
------------------------------------------------------------------------
October 8, 1991: Version 1.20 released
I had intended this release to be a bug-fix release with some
additional programs. However, classes started, and I found I
had a lot less time to work on DVSI. I was also working with
Richard Reiner on DvpEdit, a replacement for Change A Program.
DvpEdit was initially going to be released with DVSI, but we
decided to release it separately.
So anyway, this release consists of bug fixes and a few new
features.
BUG FIXES:
PS:
It was possible for negative times to be reported because
I was using long integers instead of unsigned long ints.
If a process id was specified but didn't exist, PS would
display column headings and exit. Now it displays an
error message.
/p message not properly displayed.
I mis-documented the meaning of the SYSMEM column. This
column is the remaining system memory available to the
process, NOT the total available as I had implied. If this
field is large, you can trim down the amount of system
memory in "Change A Program" to reduce the TTLMEM field.
If PS was loaded in a startup script, it was possible
for there to be 0 seconds of CPU time, which caused a
"Divide By 0" error.
DF:
Allowed drive A: and B: in drivelist.
Print "(none)" if label does not exist, rather than garbage.
/p message not properly displayed.
MEMSTAT:
Divide error was occurring if no expanded memory was present.
Who uses DV without QEMM anyway? :+)
/p message not properly displayed.
KILL:
WN KILL:
For some bizarre reason, cursor position would occasionally
get totally messed up after killing a process. I now save
and restore the cursor position so that doesn't happen.
UPTIME:
Lots of bugs introduced in 1.11. After one day, hours
just kept increasing past 24 rather than starting from 0.
Negative times were sometimes reported because signed
rather than unsigned longs were being used.
PERF:
Negative times sometimes reported.
/p message not properly displayed.
NEW FEATURES:
All programs:
Can now use "-" in place of "/" for options.
PS:
DF:
MEMSTAT:
UPTIME:
PERF:
Added /q and /t options for quiet and terse output.
Also printed to STDOUT rather than using window output
routines. This will allow the output to be parsed more
easily by BAT files.
PS:
PERF:
Changed CPU TIME and TIME fields to print in hh:mm:ss format
rather than 1/100 seconds.
DF:
Display is much more Unix-DF like. Added ability to display
JOIN/SUBST and local/network status. Can run outside of
DESQview also.
WN KILL:
Disallowed KILL command without specifying process id.
How often do you want to kill the window you are working
in with kill? Just use "exit".
------------------------------------------------------------------------
August 19, 1991: Version 1.11 released
This is essentially a maintenance release, although a few new
features have been added.
BUG FIXES:
DVSIXDI.COM:
Some users reported lockup at DV title screen. This
was because I was not checking if I got the common
memory that I allocate at DV initialization. If the user
has the minimum of 13K of common memory available,
DVSIXDI.COM would do bad things and kill the system.
In this version, if DVSIXDI finds that not enough memory
is available, it disables itself. Running any of the
programs will report the problem.
While trying to find other bugs, I found two potential,
if not already realized, problems. First, I was making
a DV API call when I shouldn't have been. This call
had to do with getting a timer. Now, I capture INT 1C
and count my own ticks, rather than have DV do it. This
changes the programmer's interface to DVSIXDI slightly.
Second, I was not saving the value of AX in some cases.
This may cause problems with software which uses INT 2F.
I moved the code to compute instantaneous CPU time from
the RESTORE_STATE call to the SAVE_STATE call. This
was done because a process's interrupts are still valid
in the SAVE_STATE call, but not in RESTORE_STATE.
WN:
Manipulating an orphaned window caused a lot of problems.
WN now checks if the window is orphaned, and doesn't let
you do a whole lot.
All programs:
When run outside DV, each program had a different way of
erroring out. I now all the programs consistent - they
simply print a message that DV 2.26 or higher is required.
NEW FEATURES:
WN:
KILL:
PS:
At user request, a leading "%" is ignored in the <process id>
field.
DF:
Added a drive list option so that only specified drives
will be reported. Also made output more Unix-like.
------------------------------------------------------------------------
August 12, 1991: Version 1.10 released
(I had it all ready to be released on August 9. However,
my hard drive controller decided to die overnight, and took
all the data on one of my drives with it. I lost 3 days of work,
and spent the weekend rewriting the 3 days of code.)
BUG FIXES
PS: Occasional crash when not using /c, because a timer was
being freed even though it wasn't allocated. All fixed now.
Some system memory wasn't be deallocated properly, now fixed
Processes started from DVTREE by Michael Weaver had
the wrong names in the NAME and KEY fields. The same
problem occurred for some users that started programs from
a startup script. I know for certain the first is solved;
I hope the second is also. It seems that DESQview
sometimes gives the wrong offset into DESQVIEW.DVO when
you ask it. I now grab the Open keys directly from
the DVP in memory and use them to search for the NAME
in DESQVIEW.DVO.
As a side effect of above, I was able to fix the bug
when Add A Program and Delete A Program were used. PS
doesn't get hopelessly confused anymore, but displays
"(not available)" in the NAME field if DESQVIEW.DVO
is changed while PS is running. They KY field is valid,
however. To fix the problem, simply exit PS and restart.
The problem will only occur if you Add A Program and
run it while PS is running.
Orphaned processes and the DV System process now
display a "?" in the SW field, rather than a meaningless
Switch Windows number.
NEW FEATURES
PS: If /d options is not specified, searches C:\DV first,
then looks at PATH to find DESQVIEW.DVO.
Reformatted fields somewhat to make more space.
Added CPU Time and CPU Utilization fields
Added /r options to dynamically resize the window
when using /c
The /c no longer clears the screen unless it absolutely
must. This got annoying to watch.
Added /l option with more information: Default keyboard,
mailbox and objectq handles, cursor and windows positions,
and window sizes.
Can specify one process id to view.
NEW PROGRAMS
DF: Free and used disk space
MEMSTAT:
Alternative to DESQview's Memory Status
PERF:
Performance statistics like CPU Utilization, etc.
WN: Kill, raise, lower, hide, suspend or top windows
UPTIME:
Displays the amount of time DESQview has been running
KILL:
Kills a DESQview process
------------------------------------------------------------------------
July 21, 1991: Version 1.0 released
Contained a very basic PS command