From: | "Trevor Talbot" <quension(at)gmail(dot)com> |
---|---|
To: | "Steve Manes" <smanes(at)magpie(dot)com> |
Cc: | "PgSQL General ML" <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: Writing most code in Stored Procedures |
Date: | 2007-08-16 01:24:38 |
Message-ID: | 90bce5730708151824t74a5f3b2s61032656673d8e1c@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On 8/15/07, Steve Manes <smanes(at)magpie(dot)com> wrote:
> I'm fairly hardcore about keeping as much business logic as I can in the
> database. In fact, I only do SELECTs from the application, and usually
> via Views. All inserts, updates and deletes are via procs. I'm a
> proponent of separating application code from presentation and
> application code from database code. And HTML from layout style, for
> that matter.
>
> In addition to the other reasons you've gotten:
> Another is because I want transactions to start and end in the database,
> not in external application code which might crash before a COMMIT.
Hmm, how do you handle this logically? Do your applications never
need to submit chunks of work at once? Or do you do something like
fill in a temporary table, and have a proc work from that?
> Another is because I'm a freelancer and generally hand off my
> applications to the client's tech department, which is often a junior
> level grasshopper who knows just enough SQL to be dangerous. Using
> stored procedures raises the bar on what they need to know about RDMBSes
> before they start hacking working code.
...although that helps explain "why" a bit more: if you're stuck with
people who can't use transactions properly, it makes sense to keep
them out of reach...
From | Date | Subject | |
---|---|---|---|
Next Message | =?EUC-KR?B?vK2x4ryu?= | 2007-08-16 01:56:54 | please! SPI_finish is strange |
Previous Message | Phoenix Kiula | 2007-08-16 01:15:44 | Re: pg_dump on local Windows, pg_restore on Linux? |