Re: Why all the max length constraints?

"Tony Andrews" <andrewst_at_onetel.com> wrote in message news:1148815655.799029.221070_at_u72g2000cwu.googlegroups.com...
> dawn wrote:
> > J M Davitt wrote:
>
> > > There are, I"m sure, tens of thousands of products that "do not
> > > implement the RM" in which field lengths are either fixed or
> > > limited to a maximum.
> >
> > Definitely. I would have expected the other to be the case too - that
> > some implementations of the RM had fixed or max lengths and others did
> > not. But they all seem to use the same strategy in this regard. Since
> > they all implement some variety of SQL, is there anything about SQL
> > that would require this approach? I'm definitely not looking for
> > something, I'm trying to understand. Thanks. --dawn
>
> dawn wrote:
> > J M Davitt wrote:
> > > There are, I"m sure, tens of thousands of products that "do not
> > > implement the RM" in which field lengths are either fixed or
> > > limited to a maximum.
> >
> > Definitely. I would have expected the other to be the case too - that
> > some implementations of the RM had fixed or max lengths and others did
> > not. But they all seem to use the same strategy in this regard. Since
> > they all implement some variety of SQL, is there anything about SQL
> > that would require this approach? I'm definitely not looking for
> > something, I'm trying to understand. Thanks. --dawn
>
> Can we assume that by now, thanks to answers from various people here,
> you will accept that "no, there is NOT anything about SQL that would
> require this approach?" Somehow I doubt it. Most (if not all) SQL
> DBMSs have an "unlimited" character type (in Oracle for example it is
> called CLOB. Of course there is an ultimate limit: something like 2GB
> IIRC).
Excuse me, but I've never used the CLOB data type, so my memory of it is limited. Isn't there some kind of limitation on CLOBs such that you can't use them in the "where" clause, or something like that? If that's the case, isn't it fair to treat CLOB as a "non relational Oracle extension"?

Also, wasn't Dawn's OP about whether there is anything about the RM that would lead implementors to impose such a restriction? That's slightly different from asking whether there is something about SQL that would lead to the same thing. Received on Wed May 31 2006 - 00:02:15 CEST