Out of the box persistence yes, embedded SQL no

[Daniel K.]

Hi,

having persistence out of the box and direct access to persisted
entities would certainly speed up development. But some of the ideas
on the start page won't convince me as they just don't fit into my
picture of a "reboot" for persistence. Two things.

1.) Please PROVIDE PERSISTENCE AT THE ACTUAL LANGUAGE LEVEL. Have you
considered the benefits of changing the development approach from
"embedded SQL" to "SQL-like functions"? Many have sacrificed the
consistency, the readability, and the maintainability of their
languages by "embedding" SQL in instead of "melting" them together.
Today, the entire java.sql package appears like one huge tutorial how
NOT to do it. IMO, every next-generation language MUST implement
persistence transparently and find means to do that on the first
language level. Hibernate etc. have done a lot that would deserve to
be picked up.

2.) Please HIDE DDL OPERATIONS FROM THE APPLICATION PROGRAMMER. In
accordance with the "duck" principle that golem.de and others mention
as one of your design priciples, I strongly recommend you rethink how
much SQL you really want in "reboot". Personally, I believe that
SELECT and the SQL DML statements (INSERT, UPDATE, DELETE) are some of
the most elegant statements ever invented. BUT at least the DDL parts
are IMO questionable. Take a look at modern key value stores. CREATE
TABLE is dead, at least on an application programmer's implementation
level. Structural extensions (new entity types, new attributes/
columns, etc.) should IMO be hidden and part of the low-level
implementation.

Regards


Daniel