On 15/01/2014 13:29, Amit Langote wrote:
> On Wed, Jan 15, 2014 at 9:02 PM, Ivan Voras <ivoras@freebsd.org> wrote:
>> On 15/01/2014 12:36, Amit Langote wrote:
>>> * In some locales strcoll() can claim that
>>> nonidentical strings are
>>> * equal. Believing that would be bad news for a
>>> number of reasons,
>>> * so we follow Perl's lead and sort "equal" strings
>>> according to
>>> * strcmp().
>>> */
>>> if (result == 0)
>>> result = strcmp(a1p, a2p);
>>
>> That seems odd and inefficient. Why would it be necessary? I would think
>> indexing (and other collation-sensitive operations) don't care what the
>> actual collation result is for arbitrary blobs of strings, as long as
>> they are stable?
>>
>
> This is the behavior since quite some time introduced by this commit
>
> http://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=656beff59033ccc5261a615802e1a85da68e8fad
Ok, the commit comment is:
"Adjust string comparison so that only bitwise-equal strings are considered
equal: if strcoll claims two strings are equal, check it with strcmp, and
sort according to strcmp if not identical. This fixes inconsistent
behavior under glibc's hu_HU locale, and probably under some other locales
as well. Also, take advantage of the now-well-defined behavior to speed up
texteq, textne, bpchareq, bpcharne: they may as well just do a bitwise
comparison and not bother with strcoll at all."
... so it's just another workaround for OS specific locale issues - to
me it looks like just another reason to use ICU.