[Testing] OLPC 10.1.2 Release Candidate 1

James Cameron quozl at laptop.org
Sun Aug 8 23:51:35 EDT 2010


On Sun, Aug 08, 2010 at 09:52:28PM -0500, Mikus Grinbergs wrote:
> > switching between the desktop
> > environments will certainly cause one of the following:
> > ... 
> > 2.  loss of association and no reassociation (the other desktop hasn't
> > been told to trust the access point).
> 
> When phrases such as "trust the access point" start being used, I have
> great difficulty understanding how "wifi with a choice of access points"
> is supposed to work.  If I want to use an access point, why does the XO
> need to be told to "trust" that access point ?

Entirely my phrasing you are criticising.  There's no terminology
defined on the neighbourhood view ... you click on an access point icon
and a connection is attempted.  In my phrasing, that's a trust.  In the
code the terminology is the creation of a Settings object for the access point,
and the object is sent to NetworkManager as an NMSettings through a
NewConnection message.  As a result, the access point is added to the
list of connections in .sugar/default/nm/connections.cfg, and a badge is
added on the screen to the access point icon, the badge is called
"emblem-favorite" (sic).  If you have a better phrasing, let me know.

> How do I tell my XO: "Trust 'Joe' more than 'Sam'" ?

No way to do that in the current Sugar 0.84 without writing additional
software.

Once Joe and Sam are trusted, badged, connected to, favourites, or used,
whatever your terminology choice, then the first one that was trusted
will be used if it is available.  If it is not, then the second one will
be used.  So you can almost reach the goal you described if you are
willing to change it to "Connect to 'Joe' before you try to connect to
'Sam'".

If you use the "Discard network history" button, all access points
become untrusted.  So you can almost reach the goal you described if you
are willing to shorten it to "Trust 'Joe' and not 'Sam'".

-- 
James Cameron
http://quozl.linux.org.au/


More information about the Testing mailing list