Thoughts on comments package

After looking at the comments package (its more than just a plugin!), I 
can see why it has a reputation for difficult installs. Clearly, adding 
additional functionality like this is an edge case with pyblosxom - 
figuring out why is good food for thought.

In the long run, I don't think that extending pyblosxom by adding more 
and more cgi scripts is the right "best practice" for this type of 
application from a dev/test standpoint. Additionally, as an end user, I 
want as little web server configuration on my plate as possible. _and_ I 
want the ability to easily use aliasing/rewriting to control how my 
url's look. The pyblosxom.cgi should be able to do what little is 
needed/required before passing on the data to the pyblosxom engine, no 
matter what functionality is being invoked.

This also requires that the pyblosxom engine be "command" driven, and 
that plugins should have the ability to implement a command method that 
can listen and respond to specific commands. The current engine 
implements an implied "view" command. An xmlrpc plugin might implement 
several new commands that conform to the various xmlrpc blogging 
standards. A feedbacks plugin would support the various 
pingback/trackback/comments commands. As Atom continues to cook along, I 
could definitely see an atom plugin implementing atom commands.

So, I'd suggest that pyblosxom support a new url pattern along the lines 
of a straightforward "$base_url/<command>/<parm>" structure. In order to 
maintain backwards compatibility, view would be the default command, so 
that a url missing the <command> structure would have the <parm> section 
sent to the default view code. However, this opens the possibility for 
naming collisions: does "$base_url/xmlrpc/test" refer to an xmlrpc 
command or my xmprpc blog topic? I'd suggest a naming convention that 
requires commands to begin with an underscore so that 
"$base_url/_xmlrpc/test" refers to a command while 
"$base_url/xmlrpc/test" refers to a topic. This would also make the 
command parse a lot easier/faster too.

It looks (to my inexperienced eye) that Pyblosxom.run() could handle 
this pretty easily, especially given its usage of the callback patterns 
already.

Thoughts? Thanks for being patient with me as I get up to speed...

--L



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click