Re: Current CVS has strange parser for float type - Mailing list pgsql-hackers

From Teodor Sigaev
Subject Re: Current CVS has strange parser for float type
Date
Msg-id 3DAFB98C.70003@stack.net
Whole thread Raw
In response to Current CVS has strange parser for float type  (Teodor Sigaev <teodor@stack.net>)
Responses Re: Current CVS has strange parser for float type  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
wow=# show lc_numeric;  lc_numeric
-------------- ru_RU.KOI8-R
(1 row)

wow=# select 5.3::float;
ERROR:  Bad float8 input format '5.3'
wow=# set lc_numeric = 'C';
SET
wow=# select 5.3::float; float8
--------    5.3
(1 row)

It's locale.

Tom Lane wrote:
> Hannu Krosing <hannu@tm.ee> writes:
> 
>>On Thu, 2002-10-17 at 23:34, Teodor Sigaev wrote:
>>
>>>wow=# select 5.3::float;
>>>ERROR:  Bad float8 input format '5.3'
>>
>>Could it be something with locales ?
> 
> 
> Oooh, bingo!  On HPUX:
> 
> regression=# select 5.3::float;
>  float8
> --------
>     5.3
> (1 row)
> 
> regression=# set lc_numeric = 'de_DE.iso88591';
> SET
> regression=# select 5.3::float;
> ERROR:  Bad float8 input format '5.3'
> 
> I think this is a consequence of the changes made a little while back
> (by Peter IIRC?) in locale handling.  It used to be that we deliberately
> did *not* allow any LC_ setting except LC_MESSAGES to actually take
> effect globally in the backend, and this sort of problem is exactly
> why.  I think we need to revert some aspects of that change.
> 
> Bruce, this is a "must fix" open item ...
> 
>             regards, tom lane
> 

-- 
Teodor Sigaev
teodor@stack.net




pgsql-hackers by date:

Previous
From: Teodor Sigaev
Date:
Subject: Re: Current CVS has strange parser for float type
Next
From: Mats Lofkvist
Date:
Subject: Re: Analysis of ganged WAL writes