Re: Reorganize collation lookup time and place - Mailing list pgsql-hackers

From Peter Geoghegan
Subject Re: Reorganize collation lookup time and place
Date
Msg-id CAH2-Wzm9xa6hKH4JhvAEVzsKr1hA5OBHkpyHL_980UP2cSK_sQ@mail.gmail.com
Whole thread Raw
In response to Re: Reorganize collation lookup time and place  (Peter Eisentraut <peter.eisentraut@2ndquadrant.com>)
List pgsql-hackers
On Fri, Feb 1, 2019 at 7:27 AM Peter Eisentraut
<peter.eisentraut@2ndquadrant.com> wrote:
> This patch will probably make a reappearance when we consider making ICU
> available as the database-wide default collation.

Somebody recently reported that an ICU locale was over 2000x faster
than the equivalent glibc locale with their query due to supporting
the abbreviated keys optimization:

https://postgr.es/m/CACd=f9cO5OcmAeBK1M5a3+t8BO7E91Ki0WaLAxFm=FBbon_HDw@mail.gmail.com

We really should get around to supporting ICU for database-wide
collations sooner rather than later. I knew that there were cases
where glibc performs abysmally, but I didn't think it could ever be a
difference of more than two orders of magnitude until recently.

-- 
Peter Geoghegan


pgsql-hackers by date:

Previous
From: Amit Kapila
Date:
Subject: Re: Is zheap on track for PostgreSQL 12.0?
Next
From: Andrew Gierth
Date:
Subject: Re: Ryu floating point output patch