GlassFish V2

From: Chad Okere (Sep 17 2007, at 19:01)

I've actually been using glassfish and the whole JPA (Java persistence API) thing (I'm not actually clear on which bits are which. There's the JPA, oracle TopLink, and Glassfish involed in there together. You code against the JPA)

<i>So this is nice and all, but...um, like...what the hell is it?

I spent 30 seconds on the URL and gave up. (Yes, 30 seconds isn't a long time, somebody's not telling its story correctly.)</i>

What it does is let you take Java objects, define a database mapping using Java Annotations or XML files and then you just call a few functions to persist your Java objects.

I was able to create an application that involved 20 or tables, something like that without writing a single line of SQL or really doing anything more then throwing a few annotations on the classes I'd already made.

One huge problem I've run into though is that trying to do huge queries causes an out of memory error. When you do a query, you get a java collection of results. In theory it wouldn't need to cache anything in memory, but it turns out that MySQL's JDBC driver caches your results by default. You have to do some incantations with the Statement object in order to turn that off, and the JPA abstracts that away, so for huge datasets, you can't do any kind of query to just list everything. (Well it may be possible to page through results, but that's kind of a pain) So for queries like that I've been doing raw JDBC.

I've run into a few other glitches that required workarounds too, but that's life.

But it's a HUGE time saver. It's easy to do much more complicated database schemas much easier then writing queries by hand.

(btw, for the project I'm working on I wrote my own parsers for Atom and RSS. Even though the Atom parser is twice as long, it was sooo much more pleasant to write. I validated against that's site's feed. )

[link]