On Dec 10, 2006, at 12:39 PM, Tim Jones wrote:

On Dec 10, 2006, at 12:36 PM, Norman Palardy wrote:

On Dec 10, 2006, at 12:31 PM, Tim Jones wrote:

Hi Folks,

Since we've move into the UB world with 2006r4, we've had reports of an oddity that occurs randomly during looped reads of call in a Shell class. The loop hangs until the user moves the mouse or causes some other system interrupt. At that event, the process continues - sometimes until completion and sometimes for just a few more iterations through the loop. As for random, the process can be called 20 times and 3 out of the 20 will exhibit the hang.
[code snipped]
Unfortunately, debugging is impossible because when it hangs, the action of moving the mouse (or clicking if the pointer's over the pause button) wakes the app up. Is anyone else witnessing odd loop hangs that continue when the mouse is moved?

There was a similar issue with r3 or r2 on Windows but I believe it's fixed. I cant say that I've seen this effect but then I'm not doing this sort of thing in 2006r4.

I'm not going to say anything about using "DoEvents" :)

So that you know I've already given that some thought - the same random hang occurs if I use a Shell mode of 0 and scrap the DoEvents loop :) .

Located and squashed. This is code that was originally written in 4.5.3 and has been ported forward with each new RB release. Back then, threads weren't nearly as useful as they are ow and I had a pair of timers that were used to force the updating of a couple of progressbars. They were apparently getting into a spinlock over who should be updating. I've removed them (since they're no longer needed), and all is well. And, the UI is actually faster as a side effect.

Just goes to show that old code should always be re-factored when using a new compiler :).

Tim
--
Tim Jones
[EMAIL PROTECTED]
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>

Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>

Reply via email to