Network Manager, Firefox and more on FC10

[Rick Stevens]

Jeff Spaleta wrote:
> On Thu, Jan 8, 2009 at 2:19 PM, Timothy Murphy <gayleard at eircom.net> wrote:
>> The reason people are confused by NM is because
>> it is almost completely undocumented.
>> If something does not work it is doubly annoying
>> to be told that "it just works".
> 
> What exactly are you looking for in terms of documentation?  Or let me
> ask it another way.
> 
> Is there any application that you can point to that uses GConf that
> you feel is well documented? NM uses GConf for configuration settings.
> is there any application that you can point to that uses gnome-keyring
> that's well documented? NM stores network credentials in
> gnome-keyring.

That would be a nice bit to have documented.  I think the issue here
is just how "integrated" NM is with the older stuff and just how do
you make it behave when it doesn't do what's expected (and that happens
quite often).  There is no documentation on, for example:

o How to ignore networks you once used but no longer want
o Which bits of the /etc/sysconfig/network-scripts/ifcfg-xxxx files it
   pays attention to
o How to ensure a network is available even before a user logs in
o Where the data is stored and how to modify/purge it

It's a nice bit of stuff, but to just say "it just works" is stretching
the truth a bit.

> Is there any service or application that uses D-Bus that you find is
> sufficiently documented?

No, nor is udev-related stuff and that's another major problem.  The
system with all these black boxes makes debugging things quite difficult
at times.

I'm not necessarily complaining, but I did offer my services to document
NM to the NM developers and Red Hat several times simply to free up
their resources to debug and further develop the thing.  My offers, so
far, have been studiously ignored.  This isn't Oz, gang, and there's
absolutely no reason to keep people ignorant of what's behind the
curtain.  A tiny bit of explanation (even a bloody man page) would do a
world of good here.

I've been a hardware and software engineer for over 35 years.  I've
never subscribed to the "build no mechanism simply if a way can be found
to make it complex and wonderful" mentality of many of my peers.
Obscurity rarely leads to good things, and just because something's
"new" doesn't necessarily make it better (is the /etc/event.d stuff any
more efficient or flexible than /etc/inittab?  I think not)

To quote Dennis Miller, "That's my opinion.  I could be wrong."
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer                      ricks at nerd.com -
- AIM/Skype: therps2        ICQ: 22643734            Yahoo: origrps2 -
-                                                                    -
-                 Dyslexics of the world: UNTIE!                     -
----------------------------------------------------------------------