On Sat, Jan 22, 2011 at 9:50 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
>> On Sun, Jan 16, 2011 at 2:28 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>>> If we were to go with this, I'd be strongly tempted to rearrange all
>>> four of the messages involved to put the operator or function name
>>> at the end, eg
>>>
>>> function 1 (oidvector[], oidvector[]) of operator family array_ops for access method gin:
btoidvectorcmp(oidvector,oidvector)
>
>> I kind of wonder if it wouldn't be even better to just *delete* that
>> from the thing altogether and write:
>
>> function 1 (oidvector[], oidvector[]) of operator family array_ops for
>> access method gin
>
>> We're trying to represent the pg_amproc entry here, and including a
>> bunch of details of the pg_proc entry to which it happens to point
>> seems almost better to be confusing the issue.
>
> Yeah, that occurred to me too. However, the CREATE OPERATOR CLASS
> syntax doesn't really draw a distinction between the referenced
> function/operator and its reference in the opclass, and I'm not sure
> users do either.
Well, I think from a user perspective the operator class machinery is
approximately clear as mud, but the CREATE OPERATOR CLASS syntax
surely seems to draw a distinction. It seems pretty clear we're
mapping a function support number, with optional types, onto a
pre-existing function.
> So I don't want to give up the details of the function
> or operator. But sticking them at the end after a colon might make it
> clearer that the func/operator is referenced by the amproc or amop
> entry, but is not the same thing.
That seems like an improvement over the status quo, so do you want to
go ahead and do that?
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
