Re: [Epiphany] load new tabs in background

[Marco Pesenti Gritti <mpeseng tin it>]

> Slightly off the subject but relevant, why not include an optional "UI study 
> module" in epiphany that people can turn off like the feedback agent in 
> mozilla?  This thing would just keep statistics on which preferences get 
> used at all and what they get set to.  That way instead of just wild 
> speculation/theory on inclusion/exclusion of configuration options, all you 
> have to do is just include an option for a release or two, see if/how it 
> gets used and then if it is like 90% one way you just fix the preference as 
> a default and remove it.  If it is 50/50 then you leave it in as an option.  
> Then ui decisions like this can be based on actual experimental data 
> (radical notion I know).  I think as long as the reason for including such a 
> module were made plain up front, most people would be really excited about 
> it.  You could even have a page on the epiphany web site which posts running 
> statistics.  That would be cool.  Aside from the option to turn this module 
> off,  the whole thing could be totally transparent to a user since it would 
> not require their interaction to send the info and the data transfered would 
> be very small.   This module could even only be enabled for pre-1.0 releases 
> or other experimental versions.

There are two reasons because I think this would not work:

- People that test pre releases are not a representative subset of our
(potential) user base, so the statistics would be wrong
- An interface is not a sum of elements, but a system. A simple "keep
just pref that most people like" is not likely to work, even with good
statistics. Statistics would need a more complex evaluation.

> Another somewhat relevant and somewhat offtopic question: why are options 
> removed from the browser and not just removed from the preferences ui so 
> that they can still be set by expert or highly motivated users using gconf?

Hidden prefs have a cost too, even if lower then visible prefs. There is
some manteinance cost and, way worst, there is the risk to not fix real
problems because there is a work around for it. (that works only for a
small subset of our userbase)

Marco