paulruizendaal wrote:

>Just a few thoughts:
>
>- So there is no CREATE/DROP SCHEMA to establish a list of permitted
>schema names? Would it make sense to use the Oracle idea of
>identifying schema's with users?
>
>

I don't like that idea at all. It was, in fact, part of Sequel 20 years
ago. The idea was that anything you created with qualified with your
name, and if you wanted to share it, you moved it to "public". It was
model so weak that almost nobody took it seriously.

I am happy with a schema default to -none- that could be changed. I
allow arbitrary length schema search lists in Netfrastructure, but I've
never used it for anything. I idea was to allow table inheritence
through an application hierarchy. Didn't prove all that useful.

>- In the future, the longer table names could help in implementing
>remote tables.
>
>

The 31 character names are from the VAX standard. Seems right foolish now.

>
>
>>It seems too simple. What am I missing?
>>
>>
>
>The two-level naming goes beyond mere relations. It should apply to
>all database objects, including sequences, stored proc's, domains,
>etc. Or should't it?
>
>

Absolutely.

>How about "myschema.table1" and "table1"? In which contexts would
>these two be the same? Is there a default or implied schema?
>
>

If the default schema were set to "myschema", then "table1" would
resolve to "myschema.table1". If that didn't exist, it would resolve to
an unqualified "table1". There is question of what happens if the
default schema is "myschema", the table "myschema.table1" exists, and
you want to reference the unqualified "table1". I can think of about a
dozens possible solutions, none the least bit interesting.


--

Jim Starkey
Netfrastructure, Inc.
978 526-1376