Re: Determining char/varchar2 column length

  • From: Tanel Põder <tanel.poder.003@xxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Thu, 15 Jul 2004 00:04:19 +0300

> That's correct, but version determines how you can calculate it to begin
> with.

> In 8i, data_length would be set to "30" (which is the byte_length), but
> if you're using a multibyte character set, the actual number of
> characters that can be stored there will vary.  (e.g., 30 US7ASCII

One minor addition, with *variable width* multibyte charsets, yes this can
vary, with fixed width it will be fixed how many bytes a char takes.

> characters, but say only 10 multibyte Japanese characters).
> In 9i, the nls_length_semantics allows one to specify how Oracle treats
> the length specified during table definition.  Using "byte" semantics,
> it acts the same as 8i.  Using "char" semantics, however, data_length no
> longer reflects the length specified in the column definition, but
> rather calculates this length (the actual number of bytes needed) using
> the database character set.  For UTF8, this is typically three-times the
> actual length specified in the column definition.

Btw, this is an important point, that nls_length_semanics affects only table
column (and possibly type) definitions, but no PL/SQL datatypes for example.
A client of mine had problems after migrating to UTF8 w. character nls
length semantics, PL/SQL variables were still defined in byte lengths..

> So, the idea is we want a query to say "here's the value specified in
> the table create statement" for the length of that column.  Whereas in
> 8i we could query data_length, in 9i data_length does /not/ always
> reflect this value.  Rather, it's stored in char_length, a field that
> does not exist in 8i.

I'm not sure whether I understand your question, but check whether


Please see the official ORACLE-L FAQ:
To unsubscribe send email to:  oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
Archives are at //
FAQ is at //

Other related posts: