Subject | Re: [firebird-support] Re: Lock conflict on no wait transaction |
---|---|
Author | David Johnson |
Post date | 2005-07-07T23:30:54Z |
I think that Helen was suggesting that you need to implement retry logic
on a transaction (rollback) due to a lock.
My solution does not eliminate the need for retry logic, or even change
the structure of that logic significantly. Primarily, the retry logic
is moved in relation to the overall code structure relative to where
Helen would have it, in a way that is counter to the general usage of
the Delphi VCL model.
I believe that Helen and I were both assuming that both the need for and
the general structure of the retry logic would be obvious in the context
of your code.
Also, your log actually indicated two writers were involved, not the
reader. I did not read your log closely enough, and I do not understand
the logs nearly as well as Helen does.
Pardon my blasphemy, but when it comes to Firebird, Ann is the Holy
Spirit and Helen is St. Peter. (Does this make Jim Jesus or God?)
on a transaction (rollback) due to a lock.
My solution does not eliminate the need for retry logic, or even change
the structure of that logic significantly. Primarily, the retry logic
is moved in relation to the overall code structure relative to where
Helen would have it, in a way that is counter to the general usage of
the Delphi VCL model.
I believe that Helen and I were both assuming that both the need for and
the general structure of the retry logic would be obvious in the context
of your code.
Also, your log actually indicated two writers were involved, not the
reader. I did not read your log closely enough, and I do not understand
the logs nearly as well as Helen does.
Pardon my blasphemy, but when it comes to Firebird, Ann is the Holy
Spirit and Helen is St. Peter. (Does this make Jim Jesus or God?)
On Fri, 2005-07-08 at 00:10 +1000, Grant Brown wrote:
> Helen,
>
> You have indicated what you think of Davids suggestion but you did not
> provide details of how this issues should be dealt with.
>
> I for one would like to see how you would deal with this issue and or
> prevent it in the first place.
>