Hmm. Hard to miss that gobs of cycles are being wasted passing both a thread
and a request object around. Since the request can only execute at any point
on a single thread, the request object can point to the active thread.
Jim Starkey
> On Apr 6, 2015, at 2:06 PM, Dimitry Sibiryakov wrote
06.04.2015 19:31, Alex Peshkoff wrote:
> You provide me your conclusion instead of what I've asked...
If you wish, here is call stack:
fbrepl.dll!ReplTransaction::processInsert(Firebird::CheckStatusWrapper
* status,
Firebird::IAttachment * att, Firebird::ITransaction * tra, Firebird:
06.04.2015 19:31, Alex Peshkoff wrote:
> If I understand you correctly you've added new type of interfaces.
No. I give up this idea and don't add new interfaces.
> When did you rerun cloop last time?
Never. It is still somewhere out of sight.
--
WBR, SD.
-
On 04/06/15 19:09, Dimitry Sibiryakov wrote:
> 06.04.2015 17:59, Alex Peshkoff wrote:
>> If you send me stack trace of what
>> happens in your case I hope I can help you better.
> There is nothing useful in call stack, only expected routines.
>
> I use a method declared like this:
>
> > #
06.04.2015 17:59, Alex Peshkoff wrote:
> If you send me stack trace of what
> happens in your case I hope I can help you better.
There is nothing useful in call stack, only expected routines.
I use a method declared like this:
> #define REPL_STATUS_TYPE IStatus
> virtual intptr_t process
On 04/06/15 18:35, Dimitry Sibiryakov wrote:
> Hello, Alex.
>
> I followed you advice and use different declaraions for a routine call and
> implementation. It stopped working now. When I use LocalStatus (or even
> CheckStatusWrapper
> made from this LocalStatus) for calling routine that w
