[Standards] publish+configure again

Maciek Niedzielski machekku at uaznia.net
Mon Apr 2 07:02:03 CDT 2007


Tomasz Sterna wrote:
> Dnia 02-04-2007, pon o godzinie 11:52 +0200, Olivier Goffart napisał(a):
>> I would say it make auto-create useless since it force you to
>> implement the 
>> complete flow 
> 
> +1
> 
> #3 is added complexity of implementations without any real gain.

#3 adds something, but maybe it's not so easy to notice at first.
Problem with #2 is that it follows "I know better" pattern. And it may
lead to unwanted results with client wanted to say "Oh, I just did it
like the last time, I had no idea something changed since then".

Let's say I'm on a confidential business trip. I change my mobile client
to publish my geoloc, etc only to a very small group of people. And then
I turn on my laptop and... p+c laptop client *knows better* what I want
to do with my geoloc and overrides my setting not even suspecting that
it may be doing something bad. Why? Because "it worked last time".
Implementing #3 enables the client to 1) know that something changed and
2) ask the user what to do about it.

I just think that it's wrong to assume that every node config change was
done by a "bad client". We should assume that the user may want to
change node's config on purpose. And it would be nice if this change
could somehow propagate to other clients, without changing it by hand in
every client.

-- 
Maciek Niedzielski
 Psi developer, XSF member
 xmpp:machekku at uaznia.net
                               " BTW: We could use PEP for this ;) "


More information about the Standards mailing list