Alaska Software Inc. - Re: Internal datastructure corrupted only in 1.9
Username: Password:
AuthorTopic: Re: Internal datastructure corrupted only in 1.9
Armando RojasRe: Internal datastructure corrupted only in 1.9
on Wed, 14 May 2008 09:33:34 -0500
I have Windows XP and I have the same problem.

My app create indexes in another Thread and... if you run twice this 
creation the next time you ZAP a file crashes the app.

any Idea for this?

Armando Rojas.

"Andreas Gehrs-Pahl" <Andreas@DDPSoftware.com> escribió en el mensaje de 
noticias news:jv7lps5tyx2a.1vd79d6ms02n1$.dlg@40tude.net...
> Simon,
>
> Sorry for the email -- wrong button!
>
>>Will crash with DbZap & DbPack.
>>If you enable the define the prog will not crash, worked fine in 1.8
>
> This works fine for me under 1.8.306 as well as 1.9.331 -- most of the 
> time.
> Your code makes the program use a lot of CPU time, which is fine as long 
> as
> nothing else happens. But if you stress the processor or HD with some 
> other
> program running (like Process Explorer), the program will slow down, and 
> you
> get an error. I ALWAYS get a simple "71:File can not be opened" on one of
> the 20 Index files. I didn't get any IDSCs at all, though. The reason for
> this error is (probably the one) described in PDR 109-5574, and it should
> have been fixed with HotFix Rollup Number 10. But I get the same "71" 
> error
> even with some or all of the HotFixes applied.
>
> I suspect that your IDSC might be related to this:
>
>>Operating system: Windows  06.00 Build 06000
>
> This means you were running the program on Windows Vista, if I'm not
> mistaken, and I don't think Alaska currently certifies (or even claims)
> that (any version of) Xbase++ is running without problems on Windows 
> Vista!
>
>>Can anybody help me!
>
> I am not quite sure what this test program should accomplish. It doesn't
> make much sense to constantly open the same file -- including 20 separate
> index files -- then to close it and to immediately open it again in a
> different thread. If you give the program some time to finish closing the
> files in the previous thread, before opening them again (exclusively) --  
> or
> even better, reduce the number of index files and/or switch to CDX files
> with multiple tags -- the program would probably cerate no errors at all.
>
> -- Andreas
>
> ---                                                                      ---
>  Andreas Gehrs-Pahl              E-Mail: GPahl@CharterMI.net
>  415 Gute Street                     or: Andreas@DDPSoftware.com
>  Owosso, MI 48867-4410               or: Andreas@Aerospace-History.net
>  Tel: (989) 723-9927           Web Site: http://www.Aerospace-History.net
> ---                                                                      ---
Andreas Gehrs-Pahl
Re: Internal datastructure corrupted only in 1.9
on Wed, 14 May 2008 23:17:11 -0400
Armando,

>I have Windows XP and I have the same problem.

>My app create indexes in another Thread and... if you run twice this 
>creation the next time you ZAP a file crashes the app.

Do you have a program that shows this behavior? What kind of error do you 
get? What DBE and which DBE Settings do you use? What version of compiler, 
Server OS and/or network do you use? How do you create the index? Do you 
first delete the existing Index file? Too many unanswered questions to 
make any informed assumption on the cause of your problems!

I have seen (very annoying) crashes due to IDSCs when using DbPack() or 
any DbFunction in another thread (like DbGoto(), etc.), but they are 
quite seldom and not reproducible in a controlled environment -- with the 
exception of (very similar) debugger-caused IDSCs. But I don't have a clue 
as to the reason or cause for any of them. 

And I know that Clayton won't like to read this, but I think that there 
was some sort of flaw introduced into the thread and/or memory handling 
with Xbase++ 1.90 that causes (file) handles to be sometimes messed up if 
multiple threads are used. Maybe some sort of buffer overflow? But Alaska 
has never explained the reason for any IDSC-type errors that they have 
fixed in the past, so it is very difficult to even guess what type of 
problem this might be and how to prevent or minimize it.

I plan on showing Steffen some of those issues at the Toronto User Group 
meeting if I'm able to attend. Maybe this will get the ball rolling.

-- Andreas

---                                                                      ---
  Andreas Gehrs-Pahl              E-Mail: GPahl@CharterMI.net
  415 Gute Street                     or: Andreas@DDPSoftware.com
  Owosso, MI 48867-4410               or: Andreas@Aerospace-History.net
  Tel: (989) 723-9927           Web Site: http://www.Aerospace-History.net
---                                                                      ---