Rhino wrote:
>>>column a VARCHAR(6000). Again, the page size you use will have to be at
>>>least 8K, otherwise the creation of the table will fail.
>>On which platform is this? On DB2 for LUW LONG VARCHAR is the precursor
>>of CLOB. It can be up to 32K big AFAIK consuming as much space as the
>>equivalent CLOB(32k) in the row).
> The question stated that he didn't expect the column to ever hold more than
> 6000 characters. That's why I proposed a definition of VARCHAR(6000).
Reread the post. You are right. I mistakingly thought you said it
required 8k pagesize for a LONG VARCHAR.

>>The difference between CLOB and LONG VARCHAR is that LONG VARCHAR is
>>"stabilized" which is one step away from deprecated.
> I haven't heard this term before. Should we be surprised if LONG VARCHAR
> disappears in Version 9 or 10? I realize that you probably can't say
> categorically that LONG VARCHAR will disappear at such and such a point
> since the planners could always change their minds....
Stabilized means that the property is not integrated into new SQL
features. Existing support remains, but it's not enhanced.
The same is true for e.g. REXX (much to the dismay of some in this group
as I know).
Deprecated is stronger than stabilized. I think DARI style procedures
are deprecated, meaning support will be removed in a future version.
I have no indication that the LONG types will be deprecated anytime soon.

Cheers
Serge
--
Serge Rielau
DB2 SQL Compiler Development
IBM Toronto Lab