Re: like/ilike improvements

"Guillaume Smet" <guillaume(dot)smet(at)gmail(dot)com> writes:

> Gregory,
>
> On 9/21/07, Gregory Stark <stark(at)enterprisedb(dot)com> wrote:
>> Hm, it does seem I missed like.c when I converted all the text operators to
>> avoid detoasting packed varlenas. I'll send a patch in a few minutes to do
>> that. I'm surprised it would have such a large effect though.
>
> The patch doesn't seem to apply cleanly on head (I have a problem with
> oracle_compat.c). I tested it though with latin1 encoding.

Huh, I'll check. You have updated recently right? Because Andrew's changes to
ascii and char and so on just went in very recently.

> The LIKE case is better:
> cityvox_latin1=# SELECT e.numeve FROM evenement e WHERE e.libgeseve
> LIKE '%hocus pocus%';
> numeve
> --------
> (0 rows)
>
> Time: 98.995 ms
>
> -> it seems to be as fast as 8.2 was, now.
>
> The ILIKE case seems to go into an infinite loop: postmaster takes
> 100% of CPU and the query never finishes.

Can you send me the test cases you're using? It seems to be working for me and
it passes all the regression tests (no idea if they test ilike though).

--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com