On 25.11.2010 18:18, Tom Lane wrote:
> Heikki Linnakangas<heikki.linnakangas@enterprisedb.com> writes:
>> I left out some details on what exactly FdwPlan should contain and what
>> it's lifecycle should be. I'm thinking that it should be allocated in
>> the CurrentMemoryContext that's active when the FDW Plan routine is
>> called, which would be the same context where we store all the Plan
>> objects. It should not be modified after creation, so that it doesn't
>> need to be copied when the ForeignScan is copied with copyObject(). It
>> should not contain transient state information like connection objects,
>> or references to a remotely prepared cursor etc. It must be possible to
>> call BeginScan multiple times with the same FdwPlan object, so that it
>> can be stored in a prepared plan that is executed multiple times.
>
> The above statements seem mutually contradictory. In particular,
> I think you're proposing that copyObject copy only a pointer and not the
> whole plan tree when copying ForeignScan.
Right.
> That is entirely
> unworkable/unacceptable: quite aside from the semantic ugliness, it will
> fail altogether for cached plans.
Hmm, I see, cached plans are planned in a shorter-lived context first,
and copied to permanent storage afterwards. Needs more thought then.
Maybe the FDW needs to provide a copyFdwPlan() function to copy FdwPlans
returned by that FDW.
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com