On 10/15/2015 05:16 PM, Josh Berkus wrote:
> On 10/15/2015 01:10 PM, Tom Lane wrote:
>> I think this means that we should get rid of proc->globals and instead
>> manufacture a new globals dict locally in each call to PLy_exec_function
>> or PLy_exec_trigger. For SETOF functions it would be necessary to keep
>> the globals dict reference somewhere in the FunctionCallInfo struct,
>> probably. Not sure about cleaning up after an error that occurs between
>> SETOF callbacks --- we might need plpython to grow an at-abort callback to
>> do decref's on unreleased dicts.
> Don't people currently specifically treat the state of the globals dict
> as a feature? That is, make use of the fact that you can store
> session-persistent data in it?
>
That was the thinking behind plperl's %_SHARED, and I assume this is
similar.
cheers
andrew