[geeklog-devel] SOCnet -- was GSoC 2010

Randy Kolenko Randy.Kolenko at nextide.ca
Wed Feb 17 11:18:54 EST 2010

By the way, finally, we're getting discussion on moving GL forward. This
is a good thing regardless of what your opinion is towards ideas.  This
is how good products are developed.

So, just further to this whole plugin (and socnet) concept:

1. I honestly could care less if socnet is part of core or a plugin.  I
strongly suggest the plugin route in order to decouple add-on
functionality from core in order to have faster development cycles and
some independence of design.

2. Based on Joe's proposal (and Joe, don't take this the wrong way), we
have to start thinking about this a bit more.  I think what's been
discussed here is far too simplistic.  
How do we handle requests to join a group?  (That would include requests
FROM a group owner to a user and vice versa.)
How do we handle blacklisting people?  How do we handle a user removing
themselves from a group for which they are not owner of? How do we
handle administration of all of this?  How do we manage privacy? How do
I manage receiving email alerts for content added for groups I am a part
All of that extra "stuff" would have to be part of core.

3. Up to this point in the GL world, plugins are mostly self-sufficient
entities.  There is next to zero collaborative abilities between them.
What if I want comments on my ABC plugin page?  Well, I have to write
comment functionality myself.  Comments on XYZ plugin?  Well, I have to
write that too.  Why not have a comment plugin that can expose itself to
other plugins via a hook based API controlled by core? Some form of PLG_
hook functionality to signal which plugin and the plugin's unique
identifier is saving, editing, inserting or deleting content that other
hooked plugins can listen for.
Turn on/off hooked plugin functionality for your product through the
online configuration for your plugin.  It's a subscription of sorts.
So, why not have a base socnet plugin and build your apps around that
dependency?  Just food for thought.  

Flame/critique away, but think architecturally and how to move GL
forward that would intrigue other developers to write code for GL.
Having 8-10 month development cycles for core would stagnate socnet and
any other plugins out there.  Socnet could easily race ahead to meet the
needs of the community without hacking core.


More information about the geeklog-devel mailing list