Remembering CTY’s traditions, it’s sacred things that can’t be explained, it’s not hard to imagine how religion arises.

BQL is Now Graph

Braindump used to access the database through a (very flaky) Object-Relational Mapper that I wrote from scratch, called DatabaseObject. I then wrote BQL, a simple query language that could perform operations on braindump data with simple statements like “get color of apple” or “set color of apple to red”, mainly for kicks — it would be cool to ‘talk’ to the system, perhaps from the command line, in something reminiscent of english.

Then I realized how useful BQL would be inside the braindump web application itself, so I replaced the entire ORM layer (and all its bugs and complexity) with BQL as the application’s main interface to the database. But embedding variables in query strings all over the place, like this, is annoying (plus ‘of’ and ‘to’ in the actual property or page name would confuse it):

BQL::query("set $property of $page_name to $value");

so I reorganized the code so each operation was in its own method, making usages look more like this (the original query language parser is still there):

BQL::set($page_name,$property,$value);

…Much better. In version 0.3, BQL also includes sorting logic:

BQL::_list(); // returns all pages
BQL::_list(‘color is red’); // only pages w/ color: red
// the _ is because PHP stupidly has a built in function called list so I can’t redefine it.
// the built in one isn’t even useful.

Anyway, in light of the actual query language not being very important (it was originally just a diversion), it’s renamed in 0.3 to Graph. Maybe GraphStore. Also, who knows, maybe this API will be useful outside of braindump.

I'm Porting Braindump to Python

I realized the biggest reason I was sticking to PHP was that it is so common on web hosts — write an app in PHP (like Wordpress, MediaWiki, or PHPBB) and anyone can run it! But no one is running braindump anyway, so sticking to PHP and missing out on all python’s awesome libraries and stuff just wasn’t worth it — it’s time to move on. Why python?

• Tons of awesome libraries — linguistics, XMPP, syntax highlighting, and everything else
• Cleaner syntax — python is about as good as it gets.
• Better design — python was designed from the ground up, PHP was evolved from its beginnings as just a couple of little functions. It shows.
• Standalone Server — it would be easy to write a standalone server program in python so anyone with python installed could download braindump and immediately run it on their computer, simply by typing python braindump/server.py into the command line and going to http://localhost:8080/ or whatever. (Right now you need a webserver and PHP — it can take a while to set up)

I’ve made a lot of changes in the PHP version since the last release (0.2), so I’ll probably release 0.3 in PHP before beginning the port in earnest.

I’ve been reading up on python web development techniques — WSGI, frameworks, templating engines, etc. Right now I’m thinking I’m gonna use Pylons as the basic framework (in turn a cobbling-together of Paste for WSGI and Routes for URL mapping), Jinja for templating, and plain SQLite for the database. (No ORM is needed because everything goes through the Graph (formerly BQL) API.

This means I can’t even run braindump on thinking-chair’s current hosting. Oh well, I wasn’t using it anyway. We could always get new hosting (although it would probably be more expensive) and then there’s always Google App Engine (w00t!).

So yeah. I’m excited.

print ‘hello python!’