Role of addons

Andrew Sutherland asutherland at
Fri Sep 14 19:14:14 UTC 2012

On 09/14/2012 11:52 AM, Kent James wrote:
>> I also wonder if some options don't necessary need UI and 
>> "about:config" is a good enough UI for them.
>   That is good for really rare choices, or cases where a minority 
> disagrees with a decision (New versus Unread counts in the Mac summary 
> is a good example of that). But some key players hate them on principle.

I think I may fall into the "on principle" bucket, so it's useful to 
enumerate the principle:  Hidden preferences can introduce combinatorial 
complexity where you need to think about and test many interacting things.

In cases where many features/preferences can potentially interact in 
complex ways, it's strongly worth avoiding them.  The 
FolderDisplayWidget/thread-pane is a big example of lots of stuff bound 
together where the presence of hidden preferences was problematic.

In cases where the features are largely orthogonal, it's a less serious 
concern, but there's still the issue that if you have a feature and it 
doesn't have tests and people don't generally know about it, then it can 
easily be regressed.

By compelling things to be extensions instead, you move the cognitive 
and testing burdens of the extra permutations off the core developers 
onto the extension developers.  This is definitely a trade-off, but can 
be beneficial to extension developers since they will need to worry 
about a simpler set of core functionality.

Having said that, I think my point is valid, but since I'm not one of 
the people moving Thunderbird forward, don't count it as a vote.


More information about the tb-planning mailing list