Hi, is this a bug or am I misunderstanding something?

db_atm=# select length(to_char(552020, '000000000'));
length
--------
10
(1 row)

in my opinion it should give 9... I am getting ugly errors for this...

Il 03/02/2010 17.25, Anselmo Canfora ha scritto:
mmm... check this one:

db_atm=# select '|' || to_char(552020, '000000000') || '|' as culptrit;
culptrit
--------------

On 02/03/2010 05:26 PM, Anselmo Canfora wrote:

I suppose it's for the '-' sign on negative numbers.

On 02/03/2010 05:25 PM, Anselmo Canfora wrote:
Yep, it's documented:

http://www.postgresql.org/docs/8.1/s...ormatting.html

FM suppresses leading zeroes and trailing blanks that would otherwise be
added to make the output of a pattern be fixed-width.

So you should use 'FM000000000'

Il 03/02/2010 17.35, Marco Mariani ha scritto:
rather strange, this behavior it is not really clear from documentation,
"that would otherwise be added to make the output of a pattern be
fixed-width" it is a part not really clear for me, in my example the
pattern is one of 9 chars, to_char is adding a blank, giving a 10 char
wide string, I don't get the reason of such implementation.

Anselmo Canfora wrote:
Distinguish between "the pattern" and "the output of a pattern".

For "to_char(-1, '0')" you get "-1", so to make the output of the
pattern fixed-with no matter what the input is, "to_char(1, '0')"
will produce " 1".

The reason for the implementation is easily given:
This is an Oracle compatibility function, so it will behave like
Oracle as much as it can.

If you think that the behaviour is silly, file an enhancement request
with Oracle and come back when they have changed it :^)

Yours,
Laurenz Albe

Il 04/02/2010 8.59, Laurenz Albe ha scritto:
now it is clear

the next time I'll see Ellison's yacht in my city harbour will jump
aboard and yell about that