[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SEUL: The Cathedral and the bazar



> The problem with SEUL is than it is adopting the Cathedral model like
> in GCC: careful maturing, hoping creating the lastest and greatest
> distribution.
I disagree.  My goal has, and always will be, to guide the project enough to 
get things done.  I do not want to control everything, as I know that Linux
and associated projects develop best when minds run free.

However, I have thought long and hard about this project, and I do not feel 
that we can adopt a 100% Bazaar structure.  The scope of the project is 
simply too large to jump in without planning at least the general structure 
and goals of the project.  We have a specific goal: to create a distribution 
for end-users.  A Bazaar model is not as useful for this simply because, left 
to their own devices, those who code software will code within their realm of 
expertise in user interface, i.e. not necessarily consistent with other 
projects or usable by end-users.

> What is needed is the bazar aproach like in Linux kernel: start
> humble, frequent releases, enthousiasm.  So let's see how to do it.

> Pick an RH5: it is not a bad starting point.
I'm leaning towards 4.2 at least check, but things may be better now that the 
errata is longer.  We could start with 5.0 (it does have many advantages) and 
keep in sync with them a little longer, say through 5.1, just to gain some of 
the stability that 5.1 is almost guaranteed to have over 5.0.

> Add a couple of packages you think they are missing and remove another
> couple thab you think they are redundant or useless for our purpose.
Such as adding in the chosen toolkit, apps, etc.

> No system is difficult if you can get help.  Include Wowen Goods in it
> (a good doc can make a difficult system, easy), find speaking German to
> translate the Introductory Kurs in Caldera Lite.
I agree.  Initially, something along these lines will be crucial.  
Eventually, though, and I think most here would agree, such things will 
become almost unnecessary, as the system does things well enough for the 
user(s) to not have to worry about administration.  But that's a ways off.

> Then announce SEUL-0.0.1 and put the system available for download.
> That can be made in about two weeks.  Then begin to replace or improve
> parts of RedHat be it parts of the installation system or components.
> Make frequent releases.  Keep moving.
This has been my goal all along.  As soon as we are able, I want to get 
something in the repository, get it building, and get something packaged for 
limited release (to the Linux developer community).  Then we can start 
building up the system into what we want.  I know that we *cannot* plan, 
plan, plan, and develop.  We have to plan, develop, develop, develop, etc.  

That's the intent of the Bazaar model.  Its goal (IMO) isn't to say that
things should be totally chaotic, it says that things need to be fluid, and 
in the hands of the masses.  GCC stopped development because a few people 
were in complete control and wouldn't put in any code they didn't write 
themselves (or so I hear).  That is the Cathedral model at work, and it has 
been proven to be flawed.

> And your users will be the people who will make most of the work either by
> combing the net and carrying you the jewels they find or by writing software
> for SEUL be it for improving install or applicative software.
This is why we need to not just develop something and throw it out there, but 
actually plan and start getting things right in the beginning.  If we develop 
something that's randomly put together (admittedly difficult to do if we 
start from RedHat), we run the risk of gaining a fairly bad reputation off 
the line, which can kill the project.  Thus, we need to keep our development 
efforts in sync enough to get that first release (0.0.1) out there in 
somewhat reasonable shape.

> This the way for making progress, not by tring to be perfect from the
> start.
I *know* we cannot be perfect from the start.  Our goal is to be "perfect" 
when we're done.  But as I have tried to impress upon everyone, this cannot 
be accomplished without at least some purpose, goals, planning, direction, 
and leadership.

I will be sending more information shortly.

     Erik Walthinsen <omega@seul.org> - SEUL Project system architect
        __
       /  \                SEUL: Simple End-User Linux -
      |    | M E G A            Creating a Linux distribution
      _\  /_                         for the home or office user