home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Monster Media 1993 #2
/
Image.iso
/
os2
/
pmview85.zip
/
BUGS.DOC
next >
Wrap
Text File
|
1993-04-26
|
6KB
|
127 lines
This file lists bugs that may impede your use of PMView. Bugs in PMView are
listed first, followed by known bugs in OS/2.
PMVIEW BUGS:
There are no known bugs in PMView 0.85. Please help us find some. We
welcome any and all bug reports at sthiagar@bronze.ucs.indiana.edu.
One big thing that looks like a bug in PMView is really a problem with
OS/2, namely:
LOADING PROBLEM:
PMView is fully multi-threaded. The thread that handles file loading is
an idle-time thread. The advantage of using an idle-time thread is that
you can open up *lots* of copies of PMView simultaneously without
slowing down your system very much. The disadvantage is that there are
some "greedy" programs out there that use so much CPU that they
essentially "block" all idle-time threads from running. If you're doing
a lengthy load operation and then start one of these greedy programs,
PMView will "freeze" until you exit the greedy program. The load status
indicator bar will get "stuck" where it is; as soon as you quit the
greedy program, the load will continue normally.
We have a workaround. If you enter
PMVIEW /NO_IDLE_LOAD
at the command prompt, then that PMView session will *not* use an
idle-time thread; rather, it will use a higher priority thread that is
harder to starve.
You can make this solution more permanent by using PMView's settings
notebook. To do this, start PMView, select _View S_ettings, go to the
Spe_cial page, and uncheck "Use _idle time loading and saving". You
must exit and restart PMView whenever you change this idle-time
setting.
If you opt to disable idle-time loading, either permanently or
temporarily, be aware that this will increase the "CPU load" on your
system, and may degrade multitasking performance.
* * *
OS/2 BUGS:
Everything from here on is a bug in OS/2 and *not* in PMView.
PROBLEMS WITH SPECIFIC OS/2 VERSIONS:
PMView works fine under OS/2 *with* the Service Pack installed, under
preinstalled version 2.00.1, and under the December (6.479) beta of
OS/2 2.1. We do not know of a *single* person who has run PMView
successfully under the original shrinkwrapped OS/2 2.0 *before* adding
the Service Pack.
Some of our beta testers have reported problems running PMView 0.85
under the March (6.498) beta of OS/2 2.1. The March beta seems to have
bugs in the screen drivers and palette manager that can corrupt the
Workplace Shell to the point that you have no choice but to shut down.
PMView can trigger these bugs, but then so can simple palette manager
test programs (such as ZOOM from my "Unofficial Guide to the Palette
Manager" article in the first issue of EDM/2.)
Of the beta testers who reported trouble running under the March beta,
some have said that the latest m21shp patch solves their problems.
Since we haven't yet gotten the latest patches, we can't comment
directly. However, you should always be careful when running beta
software, whether it's a beta of an image viewer or a beta of an
operating system.
GENERAL OS/2 BUGS:
As of this writing, OS/2 does *not* recover very gracefully if you run
out of swap space. Running *many* copies of PMView while viewing big
images may get you in trouble. (Remember, a 1024x768x256 image
represents 768K of pixel data alone!)
There's also a "feature" of OS/2's WPS that you might want to know
about. Suppose you associate *.GIF files with PMView. This means that
you'll be able to double-click on a *.GIF file (when you have a drives
object open, for instance) and PMView will start up and display that
file. The "feature" is this: If you start up PMView this way and then
do a _File _Delete, OS/2 will kill PMView shortly after the file gets
deleted. Peter is thinking of a workaround, but I don't think that
"fixing" this is a high priority. Please let me know if you disagree.
OS/2 SCREEN DRIVER BUGS:
If you run under a screen driver that does not support Palette Manager,
you won't get very good results. As of this writing, the 32-bit
256-color drivers for the XGA, ET4000, Trident, and ThinkPad 700C are
known to support Palette Manager. (These drivers are available in the
Service Pack or the latest 2.1 beta.) Also, a *very* recently-released
8514/a screen driver for the March beta is reputed to (finally) give
good Palette Manager support for that video adapter. However, the
warning about bugs in the March beta still stands.
THE UNIVERSAL BUG:
There's a bug in *every* known screen driver that supports Palette
Manager. When you drop a menu down over a window that's using Palette
Manager, the obscured part of the window will later be redrawn with the
wrong colors. This is (partly) why we have the F5 function.
THE FATAL ET4000 / TRIDENT SERVICE PACK BUG:
If you're using the Service Pack, and if you run a program (*any*
program, not just PMView) that changes the screen palette and then try
to move an icon on the OS/2 desktop, YOUR COMPUTER WILL HANG!!! Be
*very* careful if you want to try this out. This bug occurs on ET4000
systems using the Service Pack screen driver. It also occurs on Trident
systems using the Service Pack and the screen driver that was
distributed shortly afterwards. This bug is fixed in the December 2.1
beta. If you have an ET4000 or Trident, you may want to acquire the
December beta.
This bug does *not* occur on the XGA, nor on the ThinkPad 700C.
ET4000 / TRIDENT / THINKPAD SCALING BUGS:
The screen drivers for the ET4000, Trident 8900C, and ThinkPad do not
scale paletted bitmaps correctly. This is why we have our own scaling
routines built into PMView. You can disable these routines (by
selecting _View S_ettings, going to the Spe_cial page, and unchecking
the "Use PMView's _own scaling routines") to see the bugs firsthand.
LIGHT AT THE END OF THE TUNNEL:
We've heard rumors that all of these screen driver bugs have been fixed
in the most recent internal IBM builds of OS/2 2.1. With any luck, 2.1
GA will be free from these bugs, and it will be possible to remove
PMView's own scaling routines (though not until our users have updated
to 2.1!). I look forward to making this document *much* shorter.
Raja Thiagarajan / BUGS.DOC / 4-23-93 release