Subject Re: [Firebird-Architect] Atomic DDL and Metadata Transactions.
Author Jim Starkey
Lester Caine wrote:

>Dmitry Yemanov wrote:
>>>a non-standard behavior at best.
>>Could you provide a reference? I wasn't able to confirm a non-transactional
>>nature of DDL statements in the SQL spec.
>Having spent some time in the SQL2003 spec, I haven't found any real
>meat covering how to do a lot of the fundamental things :)
>Some 4000 pages of toilet paper ;)
Without even looking at the spec, I can tell you that the description of
the "create table" statement creates a table. In Firebird, "create
table" doesn't create a table, it sets up the metadata for a table to be
created (or not) sometime in the future. Looking to the specification
for something that says its ok not to create the table is a lot like
Bush's lawyer's looking though the Geneva Conventions for something that
says that it doesn't apply to a US President who thinks he was appointed
by God.

That said, if the feature is useful, well-defined, supportable, and
already in use, the standard shouldn't determine whether or not it is
implemented by Firebird.

What we're trying to do here is determine whether it is useful,
well-defined, and supportable. I think the "already in use" has been
established. [Useful is still in play because I still think that
alternative features could provide better solutions to the problem
currently solved by metadating without commitment.]


Jim Starkey
Netfrastructure, Inc.
978 526-1376