Re: system-config-dns?

["Chester R. Hosey" <Chester Hosey gianteagle com>]

On Fri, 2005-07-22 at 10:31 -0500, Ed Wilts wrote:
> On Fri, Jul 22, 2005 at 09:21:55AM -0400, Chester R. Hosey wrote:
> > On Thu, 2005-07-21 at 18:06 -0500, Ed Wilts wrote:
> > > On Thu, Jul 21, 2005 at 05:05:32PM -0400, Chester R. Hosey wrote:
> > > > If someone were sufficiently demented they could probably do something
> > > > similar with UNIX tools; provide an extensible general console
> > > > interface to which one could add plugins for Apache, BIND, Nagios, ...
> > > 
> > > http://www.webmin.com/
> > 
> > Is that to imply that the Webmin developers qualify as "sufficiently
> > demented"?
> 
> :-).  I meant to imply that webmin is an extensible general console
> interface with plugins for various tools including those that you have
> listed.
> 
> > > > > That's a cross-platform complaint.  It's perfectly valid, though.  
> > > > 
> > > > Not with MMC! It provides a unified management interface that isn't even
> > > > REMOTELY cross-platform.
> > > 
> > > By cross-platform I meant that every platform has the same issue.
> > 
> > I understood the meaning and intent of your statement -- it was more a
> > jab at Microsoft's unwillingness to export a client interface to
> > anything other than Windows.
> 
> Nope.  It's to mean that Windows doesn't really have it either. Some
> stuff is managed via mmc but most stuff isn't.  Their desktops apps all
> have differing interfaces to set the preferences, they're not very
> portable, and many 3rd-party vendors don't use MMC either.  It may be an
> extensible general console interface, but if all the products you need
> don't support it, it doesn't do you a lot of good.  
> 
> I'm not trying to blast Microsoft at all - it's just that to solve this
> problem requires more than a single vendor.

Agreed -- Microsoft has set an example with their applications, though.
It's a bit more defensible to have an interface that not everyone uses
than not to have an interface in the first place -- the ball is out of
Microsoft's court unless ISVs can provide solid reasons why MMC doesn't
meet their needs.

> Red Hat could develop the best interface for configuring services and
> applications, but if joe blow developer writes a service and wants to
> use his own custom config file, Red Has has a couple of choices, both of
> them bad:  1)  ignore the new service (what if it's something they
> really want to support?) or 2)  rewrite the portions of the app to
> support the management interface - what if the upstream developers
> determine this too Red Hat-centric and reject the changes?.

Aye. That was the direction behind my suggestion of modularizing the
configuration logic, so that there's less to maintain (new module vs.
deeper changes to the core code). You still have to convince the app
author that it's in their interest to separate the configuration from
the meat via some API, and you still have problems. However, once it's
in place it's more manageable, and the application benefits from the
separation between the configuration-setting logic and the core system
in terms of maintainability.

Of course, one could ask what Red Hat has to complain about, when they
already get to take independently developed applications as a starting
point and start selling them as part of RHEL. It's a lot of work, but it
might still be less than developing the entire application in-house.

Not knocking MS or RH. Both develop useful products that I can purchase
for less money than it would cost me to develop.

Chet