Re: Agenda for January 15

On 15/01/2014 18:44, David Singer wrote:
> I am sorry Walter, I simply don’t understand.
> 
> I am saying, if you write a browser or other UA, that accepts plugins
> or other add-ons, whose DNT header can be affected by those plug-ins
> or add-ons, you need to engineer the way that they work so that the
> rules are followed.  There are many ways to do this.

Dear David,

The problem I have with this is precisely that. The TPE would introduce
limitations to the extent a UA behaviour can be changed by add-ons. That
works perfectly fine for Safari and Internet Explorer whose licenses
allow a greater deal of control over their add-ons. It does not work for
UAs that have licenses that explicitly rule out any limitations in terms
of 'intended use'. Basically you end up imposing an obligation on UAs
not to be licensed under a license that explicitly makes the intended
use of the UA unlimited. Which throws out any open source UA.

Mozilla having a voluntary blocklist does not take away anything from
that notion. I think that blocklist already may be a contradiction with
their own MPL license. I have to look at that more closely. Regardless
of that, other open source projects may feel differently, even if they
are just a fork of Firefox. In case of any GPL UA this frankly cannot be
reconciled with the GPL.

To make myself perfectly clear: this is not an software engineering
argument, this is a legal argument. Feel free to consult your friendly
native open source licensing expert.

All of this time and energy can be spent on more useful things once you
adhere to the notion that ultimately you cannot know that the signal you
receive actually originates from the UA it claims to originate from and
that you either trust it blindly or not. Trusting it blindly makes
perfect sense in a spec that assumes the UA likewise with any signals it
receives from the server.

Regards,

 Walter

Received on Wednesday, 15 January 2014 18:25:45 UTC