Re: Progress report on locale safe LIKE indexing

Peter Eisentraut wrote:
>
> Hiroshi Inoue writes:
>
> > 1) Because the current implementaion of LIKE isn't locale-aware,
> > we should be compatible with it for ever.
>
> I'm not sure I intended to say that. The combination of the following
> factors seems important:
>
> a) compatibility is desirable
> b) no requests to the contrary have been made
> c) the LIKE locale awareness as defined by SQL is quite brain-dead
>
> > 2) strcoll(str1, str2) == 0 means strcmp(str1, str2) == 0
> > in any locale.
>
> I *think* that is true, but until something happens about 1) it doesn't
> matter.
>

Tatsuo reported a case that strcoll(str1, str2) == 0 but strcmp(
str1, str2) != 0 though it seems to be considered as an OS bug.

> What is your position: Do you think my solution is short-lived because
> the LIKE implementation is wrong and should be changed?

Yes I'm afraid of it though I'm not sure.
If your solution is short-lived, your change would be not
only useless but also harmful. So I expect locale-aware
people to confirm that we are in the right direction.
I myself don't need the locale support at all. In Japan
existent locales themselves seem to be harmful for most
people as Tatsuo mentioned already.

regards,
Hiroshi Inoue