Postgres Pro
Downloads
Home   >   mailing lists

Thread: Is this "order by" logic right

Is this "order by" logic right

From
Josué Maldonado
Date:
Hello list,

The query has the following in the where:

UPPER(cty_Name) LIKE 'NEW%' ORDER BY cty_name

it returns the data in this way:

Newark
Newark
New Britain
Newburgh
New Cannan
New Castle
Newcomerstown

The table has index on cty_name (btree), I'll appreciate any comments or
suggestions about it.

Thanks

--
Josué Maldonado.

Re: Is this "order by" logic right

From
Stephan Szabo
Date:
On Wed, 24 Sep 2003, [ISO-8859-1] Josué Maldonado wrote:

> Hello list,
>
> The query has the following in the where:
>
> UPPER(cty_Name) LIKE 'NEW%' ORDER BY cty_name
>
> it returns the data in this way:
>
> Newark
> Newark
> New Britain
> Newburgh
> New Cannan
> New Castle
> Newcomerstown
>
> The table has index on cty_name (btree), I'll appreciate any comments or
> suggestions about it.

You haven't given your locale or system type, but this seems reasonably
if you're in a locale that generally ignores spaces (for example my redhat
machine does so in en_US).  If you want byte order sorting, you'll
probably need to re-initdb with LC_COLLATE set to "C"

Re: Is this "order by" logic right

From
Josué Maldonado
Date:
Hi Stephan,

Stephan Szabo wrote:

> On Wed, 24 Sep 2003, [ISO-8859-1] Josué Maldonado wrote:
> You haven't given your locale or system type, but this seems reasonably
> if you're in a locale that generally ignores spaces (for example my redhat
> machine does so in en_US).  If you want byte order sorting, you'll
> probably need to re-initdb with LC_COLLATE set to "C"

The db was initialized with the defaults in postgresql.conf, so the
locale is en_US. Any idea how to change it?

--
Josué Maldonado.

Re: Is this "order by" logic right

From
Stephan Szabo
Date:
On Wed, 24 Sep 2003, [ISO-8859-1] Josué Maldonado wrote:

> Hi Stephan,
>
> Stephan Szabo wrote:
>
> > On Wed, 24 Sep 2003, [ISO-8859-1] Josué Maldonado wrote:
> > You haven't given your locale or system type, but this seems reasonably
> > if you're in a locale that generally ignores spaces (for example my redhat
> > machine does so in en_US).  If you want byte order sorting, you'll
> > probably need to re-initdb with LC_COLLATE set to "C"
>
> The db was initialized with the defaults in postgresql.conf, so the
> locale is en_US. Any idea how to change it?

Unfortunately, the only real way to change the collation right now is to
re-initdb with LC_COLLATE set to something else in the environment. The
index order for btree indexes is dependant on the collation order so
changing it after the fact would be bad.  This does mean dumping and
reloading or some kind of replication to get the data between the old
system and new.

Re: Is this "order by" logic right

From
Dennis Gearon
Date:
What should we comment on?

翽 wrote:

> Hello list,
>
> The query has the following in the where:
>
> UPPER(cty_Name) LIKE 'NEW%' ORDER BY cty_name
>
> it returns the data in this way:
>
> Newark
> Newark
> New Britain
> Newburgh
> New Cannan
> New Castle
> Newcomerstown
>
> The table has index on cty_name (btree), I'll appreciate any comments
> or suggestions about it.
>
> Thanks
>