Re: control center plans, etc.

[Havoc Pennington <hp redhat com>]

Dick Porter <dick acm org> writes:
> On Wed, Jun 20, 2001 at 04:08:29PM +0100, Stephen Browne wrote:
> > 
> > > - Implemented location management, some capplets are using the location
> > > manager already and it is very nice. 
> > 
> > I agree this is nice functionality, my concern is that the archiver used to 
> > accomplish this is currently part of the Ximian-Setup-Tools and I don't know
> > about other distributors but I am fairly certain that SUN has no intention of
> > delivering or supporting XST.
> > 
> > Making the Control Center depend on XST is not a really nice idea.
> > I know the archiver stuff is conditionally compiled at the minute so that
> > this dependancy doesn't truely exist, but people without XST are going to
> > loose out on the location management feature.
> 
> Isn't that a given? You don't use package A, so you lose out on package A's
> features.
> 

I think the point is that the archiver feature is reasonable for just
the control center. Chema already said he'd move it out of XST though
so I think it's a non-issue.

I want to look at the control-center stuff in more detail today and
then work out how to approach hacking on it.

(As an aside that doesn't much matter, the line between control-center
and XST is a simple one, I think Sun doesn't want to ship it for the
same reason Red Hat wouldn't - we already have our own config tools
and config backends. These are closely tied to the operating system,
and usually have to interoperate with command line tools, web GUIs,
and so on. Anyway, anytime we change anything in the OS those tools
must be updated, and in an unportable/OS-specific way. It just doesn't
make sense for us to support tools that have a lot of complexity tied
to working on HPUX or something in addition to our own OS. And it
would be hard for us to add RH-specific features; a generic framework
that has to be kept working cross-platform greatly increases the
required engineering for that. There's also the organizational issue
that the people writing RH config tools aren't involved in GNOME at
the moment, and are primarily focused on server-side stuff.

On the other hand, plain control center is changing per-user GNOME
desktop settings, and is thus not tied to the underlying OS and is a
feature we do want to ship.)

Anyway, I guess I feel obligated to also say that we haven't made XST
an "official part of GNOME" yet either - at the moment it's still an
add-on package. If we were voting on that, I might feel that XST is
kind of a waste of time because OS vendors in general just don't want
that feature, but I also might abstain from the vote since I obviously
have a Red Hat interest in the topic (basically, it's more engineering
work for us if we have to strip the XST dependency from the upstream
sources).

But again, a non-issue as far as I know.

Havoc