Charles,

<< I'm interested, as I saw this error on Friday, D10S, EDB 2.22b1. Program left doing not a lot, but with open tables in remote session, running in a VM. It does however periodically, on a timer tick, run a query to update certain info displayed in the GUI. The first thing the timer tick event does is disable the timer. >>

If you have a reproducible situation where you're seeing this and you think it's a bug, please email me a project that does so and I'll take a look.

Tim Young
Elevate Software
www.elevatesoft.com

<<Tim Young [Elevate Software] wrote:

If you have a reproducible situation where you're seeing this and you think it's a bug, please email me a project that does so and I'll take a look. >>

If we can isolate it, we will see if we can reproduce it in a small example. This is happening in a large, rather unwieldy front-end to our system which has grown somewhat over the 20 years since its Delphi 1 birth.

Charles,

<< If we can isolate it, we will see if we can reproduce it in a small example. This is happening in a large, rather unwieldy front-end to our system which has grown somewhat over the 20 years since its Delphi 1 birth. >>

I certainly understand. In the meantime, I'll also keep looking here to see if I can spot anything.

Also, if anyone else in the thread has any leads on a reproducible example, please forward on to me.

Thanks,

Tim Young
Elevate Software
www.elevatesoft.com

Tim Young [Elevate Software] wrote:
<< Also, if anyone else in the thread has any leads on a reproducible example, please forward on to me. >>

Tim, not much help I'm afraid, but we're as certain as we can be here that we had never seen a #506 before we upgraded to 2.22 recently. We're now over 3 years down the road with our rather large conversion project from DBISAM to EDB. The main part of the system has been up and running with EDB for over 2 years.

And I should add that Scott has seen this principally while debugging in the IDE.

<< Did you have to make this post - now I have received one - first ever.  >>

No need to thank me.

= Steve

Hi Matthew,

"Matthew Jones" wrote:

<< Timers can be re-entrant, so you need to check that the timer is not
already doing things. I use a simple boolean on the form for such
checks. The lock is probably better, but I mention it just in case. >>

I believe I'm using a flag but I'll double check to make sure.

= Steve

3 more customers have reported this error message.  In each case they left the application running overnight with no-one actively using it.

= Steve

Steve,

<< 3 more customers have reported this error message.  In each case they left the application running overnight with no-one actively using it. >>

I finally found the problem: it's a bug in the way the locking was changed in 2.22.  Basically, EDB is not giving the locks enough wait time to cover when a ping occurs, so any operation that randomly conflicts with the ping thread will raise this error.

I'm sending Charles a version to test, and I can do the same for you if you'd like to confirm the fix before I do a build 3.

Tim Young
Elevate Software
www.elevatesoft.com

Hi Tim,

<< I finally found the problem: it's a bug in the way the locking was changed in 2.22.  Basically, EDB is not giving the locks enough wait time to cover when a ping occurs, so any operation that randomly conflicts with the ping thread will raise this error.

I'm sending Charles a version to test, and I can do the same for you if you'd like to confirm the fix before I do a build 3.  >>

That's great news.  Sure, I'd be happy to test it.

= Steve