Re: [Setup-tool-hackers] Beginnings of NIS Support

[Hans Petter Jansson <hpj helixcode com>]

> Regarding the idea of using the oc interface for the users-conf frontend:
> 
> maybe the best possibility is for OpenConnect to have its own, custom-made
> frontend. You could use the backend, and just code the frontend.
> 
> Hans Petter: maybe that common code we are using for the frontends should
> become an installed library, so that the openconnect project requires it
> in its build system, and so then provide its own frontend in their CVS
> source tree.
> 
> I see conflicting issues between openconnect and HST, at the interface
> level, because both project address different types of users and
> configurations. Coding different frontends would give all the desired
> flexibility for the best designs. This would not be any kind of fork: it
> is a natural consecuence of HST's architecture.
> 
> Still, your help with other, new tools is very much appreciated.
> 
> I plan to release some GUI guidelines for HST starting next week, so we
> can finally discuss all these issues, starting from some base, and finally
> have a consistent, beautiful interface for all the tools.

I tend to agree with Ross in this matter.

A plethora of frontends addressing different users and different parts of the
configuration provided by the backend will

a) Have very noticeable overlaps.

b) Require the user to make an informed decision before starting to use a
   tool at all. "Am I an admin or am I a user pretending to admin?"

c) Subdivide the top-level functionality; we'll have a lot more icons in the
   control center, for instance, and many of those icons will map to tools
   that are hard to distinguish from each other conceptually.

I'm also very tempted by the idea of OC/HST being the "final word" in Unix
system configuration. The backends can handle the extra complexity (I'm
improving the code quality these days, just to be sure), and in the frontends
it's just a matter of setting and handling complexity levels. Remember, the
framework is ready for more than two complexity levels, as well.

The way I see it, HST was meant for home users because that's where initial
efforts would be most effective. I've always thought we'd extend them after
the first, and simplest, goal was achieved. If we get devoted help in doing
this right now, so much the better.

--
Hans Petter