Обсуждение: char_length()?

char_length()/octet_length() for char() datatype returns a character
length *except* the trailing blanks. Is this what the standard
expects? Oracle's length() returns 3 in the case below.

test=> create table t2 (c char(3));
CREATE
test=> insert into t2 values('c');
INSERT 277385 1
test=> select * from t2; c  
-----c  
(1 row)

test=> select char_length(c) from t2;length 
--------     1
(1 row)

test=> select octet_length(c) from t2;octet_length 
--------------           1
(1 row)
--
Tatsuo Ishii

Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> char_length()/octet_length() for char() datatype returns a character
> length *except* the trailing blanks. Is this what the standard
> expects? Oracle's length() returns 3 in the case below.

AFAICT Oracle is right --- the spec just says
             i) If the data type of S is a character data type, then the                result is the number of
charactersin the value of S.
 

and I can't see anything there about stripping pad characters.  You
could ask for length(trim(S)) if you don't want to count blanks.
        regards, tom lane

> Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> > char_length()/octet_length() for char() datatype returns a character
> > length *except* the trailing blanks. Is this what the standard
> > expects? Oracle's length() returns 3 in the case below.
> 
> AFAICT Oracle is right --- the spec just says
> 
>               i) If the data type of S is a character data type, then the
>                  result is the number of characters in the value of S.
> 
> and I can't see anything there about stripping pad characters.  You
> could ask for length(trim(S)) if you don't want to count blanks.

Ok, I have committed fixes for this. Maybe we should add this
incompatible changes to the release note:

char_length()/octet_length() for the char() datatype now returns a
character length (or byte length) including the trailing blanks.
--
Tatsuo Ishii