Australis landing soon
dolske at mozilla.com
Fri Nov 15 21:14:55 UTC 2013
On 11/15/13 9:13 AM, John Jensen wrote:
> Hi Justin,
>> We're particularly interested in add-on compatibility. Most will work
>> as-is. Some will need changes, often simple to make. A few will require
>> deeper fixes. Australis has a lot of changes, and it's difficult to
>> predict what assumptions a specific add-on has made, or how severe a
>> glitch may be. [...]
> Is there some automated / quantitative / statistical way we can mitigate these risks? Scan popular addon code for particular API calls?
Not that I'm aware of. Unfortunately it's not as simple as checking for
a certain API call. I think the risk is mostly just the result of
Australis changing a variety of things, and sometimes that can affect an
add-on in a indirect or unexpected way.
We've already done a couple add-on code surveys and checks, with mixed
Back in May we were trying to understand the impact of the addon-bar
removal, specifically regarding addons that providing wide UI widgets
(which are fine in the big wide addon bar, but don't play well in the
space-constrained navbar or menupanel). The results are not definitive
(hard to solve the halting problem with grep), but do seem to indicate
that only a handful of addons are involved.
In June, Gijs took at look at the top-100 AMO addons to see how they
fared in a migration to Australis. Of those 100, ~25 referenced the
addon-bar. Of those 25, ~15 had some sort of problem. Of those 15, ~8
were fine but had disappeared back into the customization palette (ie, a
user could drag the icon back to navbar and carry on).
We've fixed a number of issues since then (including some in that bug).
In the 5 months since we haven't had many bug reports about addons, but
neither do we know what that really means given the small/special
population currently using UX.
More information about the firefox-dev