Re: ARC & GNOME [Was: How we make decisions...]

[Mark McLoughlin <markmc redhat com>]

Hi Ed,
	I'll just make some random points :-)

  - Like Havoc said, a lot of the original reasons for the GEP process 
    have now gone away and we're faced with a different process now - 
    how does a bunch of hackers like us co-operatively make consensus 
    based decisions without an appointed decision maker (or group of 
    decision makers)

    I brought up the GEP process because I thought there were some nice 
    properties in there, despite its utter failure:

      + Encouraged people to think about the requirements for the 
        solution, before taking a final decision on the solution - 
        often the breakthroughs we have in terms of decision making
        is when we finally get a good handle on what the actual 
        requirements are - see the current polypaudio and libgnomesu 
        threads, for example.

      + It defined a group of people relevant to the topic and put the 
        responsibility on them to find consensus and make the decision.

      + It documented the discussion and decision.

  - The ARC process has similar properties, but different goals from 
    the GEP process and our current process needs. For me, the main 
    thing the ARC process achieves is the documenting of interfaces,
    their stability level and dependancies on, and duplications of,
    those interfaces.

    I think something like this is worth doing upstream, if only to
    show what the ARC means by an interface. I don't think it would
    occur to most people that the location and format of 
    ~/.gnome2/session-manual is an interface, for example. Do we 
    support applications other than gnome-session-properties parsing or 
    modifying the file? No, but we *might* consider allowing another
    control panel modify it if a true need arose. So, for now that 
    interface would be Private, but we may consider elevating it to
    Desktop Private in order to allow another control panel to use it.

    This work is worth doing, but I don't think it would be at the core 
    of the decision making process we need. Interface stability is a 
    factor in these types of decisions (see the discussion about the 
    gnome_sound_connection_get() in the recent polypaudio thread), but 
    its probably more worthwhile for the ARC to participate in defining
    the existant interfaces which would result in the entire community
    knowing what types of interface related issues to worry about when
    making decisions.

    However, for the work to be useful to the community, it would need
    to be very lightweight and done with the very latest development
    code.

Cheers,
Mark.