Page 1 of 2

Stack fault in wcbeta when opening autoscan options

PostPosted: Fri Jul 29, 2005 3:28 pm
by Glenn
Hi Derek;
I saw the new beta post and rushed with much anticipation to download the new beta and try it out. Alas, opening the auto-scan options menu crashes the program, something about a "stack fault in wcbeta" whatever that is. The program closes about 30 seconds after the ok button is clicked, leaving my pc intact. I restarted the machine but still the same message. Everything else looks ok so far. I unzipped the beta into the wavcor directory as you advised.
Anything else I could try? Thanks.
Glenn

PostPosted: Fri Jul 29, 2005 5:58 pm
by Derek
Oh dear! Not a very good start.

What operating system are you running. I've only tested under win2k and XP so it may be a OS dependant.

I can't think of anything for you to try at your end. If you let me know the exact text of the error message it might provide a clue.

I think it must be something fairly simple. So hopefully yu won't have too long to wait for a fix.

PostPosted: Fri Jul 29, 2005 7:48 pm
by Derek
I've found a problem with the initialisation of the options dialogue. I can't say for sure that it is causing your problem but it definitely needs fixing anyway. I''ve uploaded a revision (build 3.1.9.1)so if you download again, you should get the new version -assuming you can flush all your caches.

The problem arose because of a last minute change I made. (My software is like the jelly that never sets!)

Anyway, please let me know whether or not the problem is now fixed.

PostPosted: Fri Jul 29, 2005 8:05 pm
by Glenn
Hi Derek;
Ok, I've downloaded the new version and it still has what appears to be the same problem. I'm running Windows 98SE 4.10.2222A with most of the latest updates. The error message reads:
WCBETA caused a stack fault in module WCBETA.EXE at 0167:00473c2d.
Registers:
EAX=00489185 CS=0167 EIP=00473c2d EFLGS=00010246
EBX=007320a0 SS=016f ESP=00732000 EBP=0073200c
ECX=007320a0 DS=016f ESI=00000000 FS=4127
EDX=0000077a ES=016f EDI=0073206c GS=9b87
Bytes at CS:EIP:
50 c3 cc 55 8b ec 57 56 8b 75 0c 8b 4d 10 8b 7d
Stack dump:
00732108 004ac1ed ffffffff 00732038 00489862 0000bc19 00000000 0073202c 0073206c 0082f430 007320a0 0082f430 0000077a 00000006 00732060 00487c26

It crashes quite gracefully; I just need to give it about 30 seconds or so and carry on. The last version of wavcor runs perfectly, despite my occasional protestations....
Is there anything else I can provide?
Thanks again.
Glenn

PostPosted: Fri Jul 29, 2005 8:15 pm
by citguy
I also am running Win 98se. I can record normally but if I click done after recording then click "opt" to make sure filters are disabled I get an immediate crash (illegal operation message).

Stan

PostPosted: Fri Jul 29, 2005 9:08 pm
by Derek
Thanks for the reports.

I've now establihsed that the problem only affects Win98 (and probably ME). Now I can reproduce the problem, I hope I'll be able to fix t fairly easily.

PostPosted: Sat Jul 30, 2005 11:59 am
by Derek
I think the problem is now fixed. Build3.1.9.2 is now on the download page.

There was a mistake in the code that Win2000 and XP tolerated but not win98.

Thanks again for the reports.

PostPosted: Sat Jul 30, 2005 4:19 pm
by Glenn
Hi Derek;
It works ok now. So far I like what I see, but I'll need some time to evaluate it. Thanks.
Glenn

PostPosted: Sat Jul 30, 2005 6:06 pm
by citguy
Hi Derek. Just about to try the first "long project" with the beta. I could not enter my old ALPHA Registration code in the space for the registration number because it was toooo long. Am I missing something. The Beta says it is registered to me in my name. Is registration automatic?
Stan

PostPosted: Sat Jul 30, 2005 6:38 pm
by Glenn
Hi Derek;
I have another error message for you:
"A required resource was" with an ok button followed by; "not enough memory available to display the graphic", also with an ok button. This message closes alright but not the first one, and I could only close the program via the windows close program dialogue. I'm currently reviewing a recording that produced a large number of false prositives in long strings with version 3.1. I'm not seeing nearly as many this time, but the number of corrections is approaching the 20,00 limit for win98. Could this be the issue? I have 511 megs ram with a 64 meg video card and memory itself has never been an issue with wavcor, though the system monitor informs me there is less than 3 megs available when the session is done loading. I'm certain this is not unusual. This could be a potential problem as I was not able to save the session before closing the program.
I think you are on the right track here, in that you've managed to bridge the gap between a level 2 and 3 scan while keeping the false positives to a manageable level. The signature display above the main window has not been as helpful as I had hoped, and I'm wondering if the corrections display wouldn't be a better alternative. The problem is with the vertical scaling of the signature; often times the signature in the overview has better resolution until one zooms in closely enough to see any vertical deflection in the main window. Nonetheless it is still helpful.
Other than that I'm quite pleased with the changes, though I'm not sure what the minimum magnitude setting is used for. Isn't this the same as a level one or two scan?
Thanks again.
Glenn

PostPosted: Sat Jul 30, 2005 6:50 pm
by Derek
citguy wrote:Hi Derek. Just about to try the first "long project" with the beta. I could not enter my old ALPHA Registration code in the space for the registration number because it was toooo long. Am I missing something. The Beta says it is registered to me in my name. Is registration automatic?
Stan


I don't understand what you're saying. To answer your question, yes your registration will be valid for the beta. Therefore you shouldn't need to (or be able to enter your registration code. Therefore, I don't see how it could be too long. Can you explain?

PostPosted: Sat Jul 30, 2005 6:58 pm
by Derek
Glenn wrote:Hi Derek;
I have another error message for you:
"A required resource was" with an ok button followed by; "not enough memory available to display the graphic", also with an ok button. This message closes alright but not the first one, and I could only close the program via the windows close program dialogue. I'm currently reviewing a recording that produced a large number of false prositives in long strings with version 3.1. I'm not seeing nearly as many this time, but the number of corrections is approaching the 20,00 limit for win98. Could this be the issue? I have 511 megs ram with a 64 meg video card and memory itself has never been an issue with wavcor, though the system monitor informs me there is less than 3 megs available when the session is done loading. I'm certain this is not unusual. This could be a potential problem as I was not able to save the session before closing the program.


Yes, this does sound like another bug, maybe a resource leak, although I can't think of any changes that might be responsible.

Can you say exactly when tthe error message occurs and what the steps are leading up to it.

Other than that I'm quite pleased with the changes, though I'm not sure what the minimum magnitude setting is used for. Isn't this the same as a level one or two scan?
Thanks again.
Glenn

It enables you to tell the program to ignore clicks it detects if their magnitude is less than the value you specify. This is something that was asked for in the wish list. It''s another way of reducing the number of false positives as they tend to have a low magnitude compared to genuine clicks. Try it and see.[/quote]

PostPosted: Sat Jul 30, 2005 8:19 pm
by citguy
My Alpha "registration key" is 12 characters long. When I tried to enter it in the "registration box" it would only accept about 9 of the characters. But as I said, the Beta copy indicates it is registered.

Stan

PostPosted: Sat Jul 30, 2005 8:28 pm
by Glenn
Hi Derek;
The error message came up out of blue while I was scrolling the main window under close magnification to view the corrections on the new signature display. I had done a number of superscans (possibly 3), intermediately adjusting the AGC and Noise Discriminator settings to raise the number of corrections and observing them for false positives.
Incidentally, this would likely be my approach with the new version, although in the past I've prefered to do a level 2 scan and superscan at level 3 on a track-by-track basis to avoid 98se's limitation on the number of listed corrections. This issue has been a thorn in my side in the past, as it means the corrections not in the list can't be removed, as far as I know.
This new approach seems to be a much better work-around.
Glenn

PostPosted: Sat Jul 30, 2005 8:46 pm
by Glenn
Hi Derek;
I forgot to mention that while the error message was on the screen the windows metrics were being displayed incorrectly (icons and fonts in the taskbar were different). This was rectified when the program was forced closed.
Glenn