J
James Hokes
Hi, all.
Based on my experience with the .Net platform, and after reading numerous
tomes about it, I have come to the following conclusion (for right or
wrong):
The ADO.Net libraries seem designed primarily for extremely tight coupling
to the datasource. All of these adapters, datatables, relations,
strongly-typed datasets, etc. all focus on or rely indirectly on having
embedded sql in the project.
How does this tie in with the loosely-coupled architectures that I'm used to
building? How can I tie in complex stored procedures? Most of the n-tier
systems that I've been involved with have _NO_ embedded SQL in them - on
purpose; we rely heavily on stored procedures to put the logic on the server
and keep it out of the component code.
What am I missing here? I surely don't want to "drag" tables (the horror!)
onto my design surface. I don't want these apps to know anything _about_
tables, nor do I want umpteen types of database schema dependencies compiled
in. Also, I want ZERO logins on the RDBMS having direct table access. Next,
I see only very weak support for stored procedures in the majority of the
ADO.Net system. Most of it seems to be designed for the drag-and-drop
"developers", with little or no support for the hard-core, stored-procedure
based, abstracting-style of development. As far as I can tell, I can't even
get a strongly-typed dataset to represent the rows returned from a stored
procedure, unless I roll it by hand. Add in the lackluster support for NULL
data from the source and it doesn't look very appetizing at all.
Am I wrong here? I'd really like to know if I'm missing the boat on this.
Thanks,
James Hokes
Based on my experience with the .Net platform, and after reading numerous
tomes about it, I have come to the following conclusion (for right or
wrong):
The ADO.Net libraries seem designed primarily for extremely tight coupling
to the datasource. All of these adapters, datatables, relations,
strongly-typed datasets, etc. all focus on or rely indirectly on having
embedded sql in the project.
How does this tie in with the loosely-coupled architectures that I'm used to
building? How can I tie in complex stored procedures? Most of the n-tier
systems that I've been involved with have _NO_ embedded SQL in them - on
purpose; we rely heavily on stored procedures to put the logic on the server
and keep it out of the component code.
What am I missing here? I surely don't want to "drag" tables (the horror!)
onto my design surface. I don't want these apps to know anything _about_
tables, nor do I want umpteen types of database schema dependencies compiled
in. Also, I want ZERO logins on the RDBMS having direct table access. Next,
I see only very weak support for stored procedures in the majority of the
ADO.Net system. Most of it seems to be designed for the drag-and-drop
"developers", with little or no support for the hard-core, stored-procedure
based, abstracting-style of development. As far as I can tell, I can't even
get a strongly-typed dataset to represent the rows returned from a stored
procedure, unless I roll it by hand. Add in the lackluster support for NULL
data from the source and it doesn't look very appetizing at all.
Am I wrong here? I'd really like to know if I'm missing the boat on this.
Thanks,
James Hokes