Re: Fwd: [HACKERS] int type problem in 7.3

2002-10-02 Thread scott.marlowe
I'll use >indexing when I rewrite my queries to: > select 1 from mytable where id::text='' > > Regards, > Mario Weilguni > > -- Weitergeleitete Nachricht -- > > Subject: [HACKERS] int type problem in 7.3 > Date: Wed, 2 Oct 2002 08:31:45 +

Re: Fwd: [HACKERS] int type problem in 7.3

2002-10-02 Thread Nigel J. Andrews
On Wed, 2 Oct 2002, Mario Weilguni wrote: > >> But oracle accepts this one: > >> SQL> select * from re_eintraege where id=''; > >> no rows selected > >> because oracle treats the empty string as NULL > > > >Oracle does that for string data, but it doesn't do it for numerics > >does it? In any ca

Re: Fwd: [HACKERS] int type problem in 7.3

2002-10-02 Thread Mario Weilguni
>> But oracle accepts this one: >> SQL> select * from re_eintraege where id=''; >> no rows selected >> because oracle treats the empty string as NULL > >Oracle does that for string data, but it doesn't do it for numerics >does it? In any case, that behavior is surely non-compliant with >the SQL s

Re: Fwd: [HACKERS] int type problem in 7.3

2002-10-02 Thread Tom Lane
Mario Weilguni <[EMAIL PROTECTED]> writes: > Ok, I checked this again. Up until 7.2, it was possible to compare an empty string >to a number, and it worked:: > e.g.: select * from mytable where int4id='' > worked fine, but delivered no result. No, that was not what it did: in reality, the '' wa

Fwd: [HACKERS] int type problem in 7.3

2002-10-01 Thread Mario Weilguni
I'll use indexing when I rewrite my queries to: select 1 from mytable where id::text='' Regards, Mario Weilguni ------ Weitergeleitete Nachricht -- Subject: [HACKERS] int type problem in 7.3 Date: Wed, 2 Oct 2002 08:31:45 +0200 From: Mario Weilguni &

[HACKERS] int type problem in 7.3

2002-10-01 Thread Mario Weilguni
It seems queries like: select ... from table where id='' (an empty string) do not work anymore, it worked up to 7.2. This will make migration to 7.3 quite difficult for some application, especially for oracle applications. Would'nt it be better to evaluate such expressions to false. Regards,