Hello
2008/5/3 Tom Lane <tgl@sss.pgh.pa.us>:
> "Pavel Stehule" <pavel.stehule@gmail.com> writes:
>> 2008/5/2 Heikki Linnakangas <heikki@enterprisedb.com>:
>>> How about taking a completely different strategy, and implement the
>>> CASE-WHEN construct fully natively in plpgsql, instead of trying to convert
>>> it to a single SQL CASE-WHEN expression? It's not a very good match anyway;
>
>> It was first variant. It's simpler for parsing and slower for
>> execution :(. It means more than once expression evaluation and for
>> simple case value casting and comparation.
>
> I agree with Heikki: this patch is seriously ugly, and "slower for
> execution" isn't a good enough reason for saddling us with having
> to maintain such a kluge in the parser.
>
> I don't really see why you should need to have multiple expression
> evaluations, anyhow. Can't you evaluate the test expression once
> and inject its value into the comparisons using CaseTestExpr,
> the same way the core CASE-expression code works?
>
>
I have to look on this way.
Regards
Pavel Stehule
regards, tom lane
>