Re: ISO8859_1 vs UTF-8 Performance?

On 10/22/2013 6:07 PM, Caroline Beltran wrote:
> John, thank you for taking the time to respond. Performance wise, a 5%
> difference is much better than I thought possible.
>
> The only concern in regards to my post would be in regards to accent
> character collation, i.e.:
>
> Angel Smith
> Ángel Smith
> Angel Williams
> ...
>
> Additionally, accent insensitive searching is also useful because
> people may not type in the accented character when searching.  Is it
> possible to create a user defined function can be created and then
> used during index creation?  If so, can that index be used to sort
> your data display as well as for your SELECT statements?

the built in lower() (or upper() functions work with the database's
defined LC_CTYPE setting, and properly convert upper/lower case.

but, yes, it IS possible to use user defined functions for a functional
index, just not required in this case as the required functionality is
built in.


And, as I said, there is also the contributed "CITEXT" extension,
distributed with the postgres core,
http://www.nytimes.com/interactive/2012/02/12/us/entitlement-map.html?ref=us
which further simplifies this.



--
john r pierce                                      37N 122W
somewhere on the middle of the left coast