Subject | Re: [IB-Conversions] Re: Key size too long (from Oracle to Interbase) |
---|---|
Author | Helen Borrie |
Post date | 2000-08-17T07:54:28Z |
At 07:39 AM 17-08-00 +0000, you wrote:
think it is possibly a bit too short for 3 or 4-byte character keys. I
have a system which stores data in Unicode at column level but every piece
of Unicode data is uniquely identifiable by integer or small char()
identifiers made of 1-byte characters.
If your question "What am I supposed to do..." was not rhetorical, you
might like to consider a changeover as a good time to "fix up" sub-optimal
database design features like using chunky, meaningful data columns as
keys, especially if your key relationships involve multiple columns and you
are involving them in joins and subselects. These features are usually a
legacy from desktop database systems and are *not* recommended in systems
that provide optimisation - given that optimisation depends on indexes...
Still, it's not a quick-sharp solution, so I guess the question was rhetorical.
Helen
http://www.interbase2000.org
___________________________________________________
"Ask not what your free, open-source database can do for you,
but what you can do for your free, open-source database."
(J.F.K.)
>HelloActually, as Claudio correctly pointed out, the limit is 252 bytes. I
>
>Thanks to Claudio and Helen for their reply ... Hmmm ... I don't like
>very much to have a limit of 255 bytes on the length of my keys ...
>What am I suppose to do if I need to have and combined index for a
>two or three fields that are varchar(100) each ? Change my design and
>screw up my code? It's not very good don't you think? Anyway, this is
>just the voice of my frustration ... grrr...Thanks again for the
>answers.
>
>/B
think it is possibly a bit too short for 3 or 4-byte character keys. I
have a system which stores data in Unicode at column level but every piece
of Unicode data is uniquely identifiable by integer or small char()
identifiers made of 1-byte characters.
If your question "What am I supposed to do..." was not rhetorical, you
might like to consider a changeover as a good time to "fix up" sub-optimal
database design features like using chunky, meaningful data columns as
keys, especially if your key relationships involve multiple columns and you
are involving them in joins and subselects. These features are usually a
legacy from desktop database systems and are *not* recommended in systems
that provide optimisation - given that optimisation depends on indexes...
Still, it's not a quick-sharp solution, so I guess the question was rhetorical.
Helen
http://www.interbase2000.org
___________________________________________________
"Ask not what your free, open-source database can do for you,
but what you can do for your free, open-source database."
(J.F.K.)