Proposal for an extended callbacks field

[Gary Kramlich]

Eric Richie wrote:

<snip>

> This is why I log everything...

<snip>

> -Eric Richie

I've been trying to follow this thread and find a place to respond, but 
yeah, I've lost my patience and concentration :)

Anyways, I see this as a chicken and egg problem.  If we were 
gobjectified at this point (hey, i'm not pointing fingers since i'd be 
pointing at myself), we could just create some interfaces for all of 
this.  Thus we could have something like a ProtocolPluginRegistration 
interface that could have register and unregister.  Admittedly, this 
would be less optimal then properly adding them to the ProtocolPlugin 
interface (something I have in the works in my local db or working copy, 
or had somewhere).  But it would at least not require a major version 
bump.  It'd be a minor, and we could open a ticket or what have you for 
the next major to clean this up.

Also, shoot me for mentioning this if it upsets you (and yes it's 
hacky), but there is absolutely NOTHING stopping anyone from declaring 
another static struct, and placing a pointer to it in the protocol data 
on the connection.  Of course that probably isn't optimal, but is 
another course.

Anyways, these were just some random thoughts I had while reading this.

-- 
Gary Kramlich <grim at reaperworld.com>