Much work will be necessary to get Bricolage 2.0 up and running. Currently, David and I have managed to put together a schedule based upon our actual development speed versus our estimated development speed and we think it's possible that we can get a basic system running by then. Actually, all it will actually do is allow a user to authenticate and change their password via a Web interface, but much of the underlying work will already be in place and development after that should be fairly rapid.
Now to see if our projections and reality are cognizant of one another.
Re:feature highlights?
Ovid on 2005-04-13T16:18:37
You'd have to corner David for that. I'm not sure which specifics he's released. However, I can describe things that I know are public knowledge. The object store will be one of the highlights, even though the customers won't see it. We have what is essentially a reflection based Perl object system built on top of Class::Meta, but even though we're currently targeting relational databases, there is no requirement to do so. We could target multi-value databases, XML datases, LDAP stores or even flat files, if we wanted.
At the database layer, (which David will be describing at OSCON) the underlying system is like nothing I've seen in Perl and, to be honest, like nothing I've seen anywhere else. However, it neatly solves many issues with the object-relational impedance mismatch. This means we have to back off our initial desire to support MySQL because it's still too primitive for our needs, but I'm simultaneously building data store for PostgreSQL and SQLite to ensure that we can meet a design goal of being able to support any data store that matches the API spec. Incidentally, that was part of my frustration expressed in my recent post about lexing and parsing.
As a side note, I find it rather amusing that SQLite is robust enough for our needs when MySQL isn't. Unless it's changed in the last release, MySQL's updatable views only worked if you were only updating one underlying table. That's not quite useless, but close enough. You can still target MySQL with our system, but it's more work than we wanted to do right now.
Re:feature highlights?
perrin on 2005-04-13T16:33:24
I'll be sure to see David's talk about the database stuff. I try to keep up with the ideas being used in various O/R tools like Hibernate and ActiveRecord. Your mini-language looks a bit like what Tangram uses for querying.
I've never really liked database views. They feel like a hack to me, i.e. just a way to avoid writing out the actual SQL for what you're doing. Security (granting limited access to specific users) is the only use for them I've seen that I liked. I'll be interested to see what you guys have done with them though.
