Thanks Cory,
Yah, the idea would be to refactor the logic out, and pass in a version strategy. The 1.0 code would contain the same logic, but the 1.1 code would have a few more smarts, like sending out a getCount call (notice, this may not always be required ... in particular if there is an active transaction, then the fid set would be faster).
Cheers,
DZ
On 2/1/06, Cory Horner <[EMAIL PROTECTED]> wrote:
David Zwiers wrote:
> Jody is correct, we decided to not implement this (unless there is a
> real transaction), because the only posible implementation for 1.0 is
> a feature-scan ... so left it up to the user.
>
> I don't think we defined an action for AUTO_COMMIT, but -1 sounds
> correct as a default value for unknown.
>
> I can take a look at this, or you can add the two lines (will need to
> do an instanceof check for the
> TransactionStateDiff, and when there is a failure return a -1 count).
> It will work for real transactions, because at that time we have a
> fid-set, which is countable.
Thanks guys, I think I follow -- ideally we will create optimizations
for the newer versions of WFS at some point?
Please see r17840 -- i've added the instanceof check you mentioned.
Cheers,
Cory.
