Re: WG: Packages: Again

Wolfgang Wilhelm wrote
> - The more difficult a database change including rewriting of code will
> get the less likely you'll find something paying for it. In my case there
> is a list of reasons from the customer _not_ to switch from Oracle to
> PostgreSQL. Besides more obvious reasons like APEX applications on the
> list there are things like "complicated PL/SQL code e.g. ... packages..."
> (whatever complicated is). Lots of the other reasons on that list begin to
> blur because of the changes of the recent versions or the near future like
> parallelisation or working on partitions.
> Of course there are some questions about style, maintainability... But
> this would be another post.

We are a similar shop: mostly Oracle and increasingly more Postgres.

But we essentially stopped (or are in the process of) using packages
altogether - /because/ of maintainability. If a package contains more then
just a single procedure it's impossible for two devs to work on different
procedures because the package body still needs to be a *single* source file
(which sometimes means: a single file with 10 or 20 procedures). Wherever we
have the chance we started migrating packages into standalone procedures.

Which is a bit cumbersome given Oracle's limit on 30 characters for
identifiers - but it still increases maintainability. And one of the
advantages given for packages was the increase in namespace availability
which is much easier with Postgres anyway.

Just my 0.02€

--
View this message in context: http://postgresql.nabble.com/Packages-Again-tp5938583p5938892.html
Sent from the PostgreSQL - hackers mailing list archive at Nabble.com.