Re: Re: [BUGS] Unnexpected results using to_number() - Mailing list pgsql-hackers

From Karel Zak
Subject Re: Re: [BUGS] Unnexpected results using to_number()
Date
Msg-id Pine.LNX.3.96.1000710091930.24431C-100000@ara.zf.jcu.cz
Whole thread Raw
In response to Re: Re: [BUGS] Unnexpected results using to_number()  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: Re: [BUGS] Unnexpected results using to_number()
List pgsql-hackers
On Mon, 10 Jul 2000, Tom Lane wrote:

> Karel Zak <zakkr@zf.jcu.cz> writes:
> > On Sun, 9 Jul 2000, Tom Lane wrote:
> >> "Andrew Snow" <andrew@modulus.org> writes:
> >>>> # SELECT to_number('12,454.8-', '');
> >>>> pqReadData() -- backend closed the channel unexpectedly.
> >> 
> >> In current sources I get a NULL result, which seems to be what the
> >> code author intended originally.  However this seems a little bit
> 
> >  my original code not return NULL, but return numeric_in(NULL, 0, 0) for
> > this situation.
> 
> Yeah, I know.  What did you expect that to produce, if not a NULL?
It is a numeric_in() problem :-), but yes, it is still NULL.

> 
> >> inconsistent --- shouldn't it raise a bad-format error instead?
> 
> What do you think about raising an error instead of returning NULL?

Oracle:
SVRMGR> select to_number('12,454.8-', '') from dual;
TO_NUMBER(
----------
ORA-01722: invalid number


I mean that we can use ERROR here too. My original idea was same form for 
to_char and for to_number --- for to_char() Oracle say:

SVRMGR> select to_char(SYSDATE, '') from dual;
TO_CHAR(S
---------

1 row selected.


I not sure here what is better. If you mean that ERROR is better I will 
change it in some next patch fot formattin.c.
Comments?                    Karel





pgsql-hackers by date:

Previous
From: Karel Zak
Date:
Subject: Re: Per-database/schema settings
Next
From: Karel Zak
Date:
Subject: Re: Per-database/schema settings