Bug: active program closes with MA2

[Edgar]

Dale,
if you want to do another test, try running Goldfish 2.0 or Sharks 2.0 to see if that has the problem. Both of those are DirectX 7 like the Carousel.
However, Carousel may have fixed your problem because it is acting as a main window buffer to be closed. Basically Carousel is intercepting the extra window close message that your system is sending.
If Goldfish 2.0 and Sharks 2.0 is working on your system then the fix will require migrating MA2.6 to DirectX 7 or higher.

[Dale]

Originally posted by Edgar:

Dale,
if you want to do another test, try running Goldfish 2.0 or Sharks 2.0 to see if that has the problem. Both of those are DirectX 7 like the Carousel.
However, Carousel may have fixed your problem because it is acting as a main window buffer to be closed. Basically Carousel is intercepting the extra window close message that your system is sending.
If Goldfish 2.0 and Sharks 2.0 is working on your system then the fix will require migrating MA2.6 to DirectX 7 or higher.  

OK, I am now running Goldfish 2.0 (without Carousel, of course).

I'm sure it would be valuable for folks who are running MA 2.0 or 2.6 and who are seeing the close-application problem, to try using Prolific's Carousel, and report whether they are still seeing the problem.

[FerdFerd]

I haven't dropped out here. Over the past 24 hours I've thought that I had no more problems after updating my video and mouse drivers, but I am now fairly sure I've had two times out of dozens where the active application closed when I've exited the screen saver with a mouse move. I'm not sure what I should try next - Carousel or Goldfish. Probably Goldfish. I'll report back in a day or so.

[Tiny Turtle]

FF,
Great. Please do so. We're awainting your conclusions.
/TT

[FerdFerd]

I installed Sharks and Goldfish. Don't care for Goldfish, so I'm testing with Sharks. No problems yet, but it will take time to be sure.

Is there any reason to believe the nag screen that comes up on exit would affect whether other active applications close on exit?

[Dale]

Originally posted by Dale:

OK, I am now running Goldfish 2.0 (without Carousel, of course).

I'm sure it would be valuable for folks who are running MA 2.0 or 2.6 and who are seeing the close-application problem, to try using Prolific's Carousel, and report whether they are still seeing the problem.  

Approx. 24 hours running Goldfish 2.0, with no problems seen at all (except that Goldfish 2.0 demo really inhales briskly). I really don't like sharks, so I'm not going to try that one. I'll keep running Goldfish for another day or so.

I would like to see a few more data points, from other folks who are having the problem. However, it _appears_ that we have a rather conclusive theory that it's a fundamental problem in MA, because of the DirectX 6 stuff.

I guess we should have all taken the time to do all of this testing with MA 2.0.

Edgar, do you have any guess about when the necessary fixes to the MA internals might be made? [Not asking for a commitment - just a guess]

[FerdFerd]

In two days of testing with Sharks, I have not seen any examples of active applications closing when exiting the screen saver with a mouse move. I can't yet say with confidence that there will never be any, but so far, none.

One caution that occurs to me is that the trial version has a nag screen that appears on exit. Depending on how that might influence behavior on exit, it is conceivable to me that the nag screen intercepts whatever information that causes other active applicions to close. If that is the case, then this isn't a very useful test. The authors will know better than I whether this is an issue of any concern.

[Dale]

Originally posted by FerdFerd:

In two days of testing with Sharks, I have not seen any examples of active applications closing when exiting the screen saver with a mouse move. I can't yet say with confidence that there will never be any, but so far, none.

One caution that occurs to me is that the trial version has a nag screen that appears on exit. Depending on how that might influence behavior on exit, it is conveceivable to me that the nag screen intercepts whatever information that causes other active applicions to close. If that is the case, then this isn't a very useful test. The authors will know better than I whether this is an issue of any concern.  

I had the same concern about the Goldfish screensaver. I ran it for 2 days with no evidence of the problem, but it does have that nag screen that might be a concern. Given that I don't like it as a screensaver, I'm not likely to pay money just to test it for this purpose.

I have now switched to 3DWindowsXP for a day or two. I don't really like it either, but it doesn't have a nag screen.

[Dale]

I've been testing MA2.6 using build 439 (1,745,264 bytes), which I downloaded on February 22.

Would build 475 (1,744,037 bytes) which was made available recently (today?) have any affect on the "close active program" syndrome? Or is that for something else entirely?

I noted that the .exe for the new one is actually smaller.

[Edgar]

The new version was only to fix the 8th fish.
There is no code change to fix the closing of other applications because I have no clue yet how to fix that.
The nag screen from the Goldfish and Sharks are more likely helping on the problem. These other screensavers hasn't shown yet that the newer versions fixes the bug.
Note that I still think the problem is not DirectX version. I think some other application is either sending a "close message" or is changing the "close message" sent by the screensaver.
The Carousel works because it acts as Nag type of buffer though it doesn't show a Nag screen.
Dale and FerdFerd, let us know the results on the testing with Goldfish and Sharks once you have your keycodes from Prolific.

[Dale]

Originally posted by Edgar:

The new version was only to fix the 8th fish.
There is no code change to fix the closing of other applications because I have no clue yet how to fix that.
The nag screen from the Goldfish and Sharks are more likely helping on the problem. These other screensavers hasn't shown yet that the newer versions fixes the bug.
Note that I still think the problem is not DirectX version. I think some other application is either sending a "close message" or is changing the "close message" sent by the screensaver.
The Carousel works because it acts as Nag type of buffer though it doesn't show a Nag screen.
Dale and FerdFerd, let us know the results on the testing with Goldfish and Sharks once you have your keycodes from Prolific.  

OK, First I'll report that 3DwindowsXP screensaver has not exhibited the problem in about 1 day of testing.

However, from the above, I infer that results with 3DwindowsXP is not significant. So, I'll change to Goldfish right now.

I haven't had this problem with anything except MA (2.0 and 2.6) yet. So, it appears that the test results for me will be either:

A. Problem only ocurs with MA; or
B. Problem occurs with MA, Goldfish 2.0, and Sharks.

If you conclude that "I think some other application is either sending a "close message" or is changing the "close message" sent by the screensaver. " - and the evidence is that it only affects MA, Goldfish, and Sharks, and not 3DwindowsXP (or any other screensaver I've tested), then what?

Of course, that's hypothetical.

[Dale]

Approx. 24 hours using Goldfish 2.0, with absolutely no observed problems.

I'll change to Sharks for the next day.

[Edgar]

Thanks for helping out Dale. I am glad Goldfish is working for you. It is beginning to look like DirectX 6 is not compatible with your system. We will wait for FerdFerd and see if he has the same result.

[FerdFerd]

Originally posted by Edgar:

We will wait for FerdFerd and see if he has the same result.  

I didn't get to start the real test until last night. I'll run Sharks over the weekend and report what I learn. If I have an incident where exiting causes an application to close, I'll report that right away.

[Dale]

Originally posted by Edgar:

Thanks for helping out Dale. I am glad Goldfish is working for you. It is beginning to look like DirectX 6 is not compatible with your system. We will wait for FerdFerd and see if he has the same result.  

Is there some way that we can directly test that premise?

[Edgar]

Dale, What do you suggest?
About programming time, I work on whatever project that Prolific's highest priority is. We have many projects that we work on. I am pretty much doing this support on my own time and that is why I am available on weekends and odd times.
By the way, is there any drawbacks to using Carousel as a fix for MA2.6 for now?
I am sure when the 3D background version that Jim is working on comes out, it will use the newer DirectX. Of course we don't know when that would be out.
Actually there was something for Jim that I should be doing but haven't had the time.

[Jim Sachs]

Ha, ha - Edgar is referring to some experiments and demos using Shaders that he and Will Ware were going to do for me. I've pretty much given up on Shaders and am attempting to do the 3D background "old school".

[Dale]

Originally posted by Edgar:

Dale, What do you suggest?
About programming time, I work on whatever project that Prolific's highest priority is. We have many projects that we work on. I am pretty much doing this support on my own time and that is why I am available on weekends and odd times.
By the way, is there any drawbacks to using Carousel as a fix for MA2.6 for now?
I am sure when the 3D background version that Jim is working on comes out, it will use the newer DirectX. Of course we don't know when that would be out.
Actually there was something for Jim that I should be doing but haven't had the time.  

I certainly don't have any ideas about how to diagnose the specific problem with MA - we're obviously looking at it via circumstantial evidence. If you can't think of a direct way to test, that's what we're left with. Just demonstrate that nothing except MA (2.0 and 2.6) causes the problem.

The path forward would then be to change MA2.6 to get rid of DirectX 6 stuff, and demonstrate that doing so eliminates the problem - OR DOES NOT DO SO.

Now, you ask some other great questions...

Well, let me give some reactions - I hope this isn't taken the wrong way. And I hope you don't take it personally.

First - thanks for working on this on your own time, weekends, odd times, etc. I appreciate that.

However, I'm disappointed that Prolific apparently doesn't consider finding and fixing a bug that's been in MA since 2.0 or before, to be any sort of a priority.

Of course, I'm also working on this on my own time, weekends, odd times, etc. I would rather be doing something else, instead of changing screensavers, etc. etc.

But I had the idea that I was doing something productive for the community, that would lead to fixing a bothersome problem. Perhaps a problem that was causing other customers to silently turn off MA, and tell all their friends about it. That is - helping Prolific/SereneScreen and the MA enthusiasts.

I think that I (and others here) have done what we could to help you (SereneScreen) understand the problem. I would suggest that many of us had some expectation that our help would lead to fixing the bug.

I guess there has been some help. We seem to know that running Carousel on top of MA is a legitimate workaround. That presumably helps SereneScreen tech support, and will help some users (especially after it appears in the FAQ). If/when those users hunt for help.

There's no big drawback for *ME* to use Carousel - but of course, that's because I know about it.

Finally, you indicate that "when the 3D background version that Jim is working on comes out, it will use the newer DirectX". I would have expected the same for MA2.6. So, you're implying that there won't be any real corrective surgery for the "close application" problem until the next major version of MA.

Because we can't diagnose this problem with MA2.6, and no more work will be done on it, there is some chance that the problem will not be fixed by using "the newer DirectX".

I'm guessing (based on past and current evidence) that if the problem is NOT fixed in the next version, then it will still not be a priority for SereneScreen.

So, I suppose I should say "how would you expect me to feel"?

Or, to ask you the same question you asked me: Edgar, What do you suggest?

Is it fair for me to ask that?

[Jim Sachs]

Dale - make no mistake, your help is greatly appreciated. Your scientific approach is quite refreshing. You are dealing with a guy (Edgar) who is extremely experienced in tracking down bugs. This one is so rare that no one directly connected with the program has ever seen it, so he's flying blind in this case. What makes it harder is the fact that it seems to be tied to DirectX, a Microsoft program that we do not have the source code for. Our code doesn't do anything weird - the program opens, moves some graphics around, and closes. Just like every other program either I or Prolific have ever written. All the C calls are done in the normal recommended way.

Switching from a DX6 base to a DX7 base is way beyond the scope of this little 2-fish update. MA3 will use DX8 or later, and that's where our priorities lie. (But, yes - it would really be nice to fix this bug in the current version.)

[Edgar]

Dale, Please read this link.
http://forum.networktechs.com/archiv...819071031.html

This person also had this same problem but not on MA2.6 screensaver.

They found a solution but I don't know if it will apply to you.

Since you have an Intellipoint Mouse, can you try using the normal PS2 Mouse driver to see if your problem disappears?

Note: On a bug that only happens on very few people, the best debugging procedure is to boot the same hardware with a clean system then install the suspected software by itself. If it fails then that suspected software is to more likely to blame. If it does not fail, start installing other software until the same problem occurs. The last software installed when the problem occurs is more likely the cause of the problem.

Since I don't have access to the same hardware, this is not possible for me.

Like I said before, I still don't think the problem is caused by the MA2.6 screensaver alone, but another software (maybe video driver, mouse driver, other currently running software) may have conflicts with it.

Edit: I am troubleshooting the bug with you to see if there is a way for me to work around it with MA2.6, however not knowing exactly what is needed to fix a problem is directly related to the priorities of when it gets done.