Proposal Overview:
Firefox's switch to rapid releases has been stressful for add-on developers. Add-ons not hosted on AMO are especially pained, as they must update compatibility every 6 weeks without the benefit of automatic compatibility bumping.
Since the release of Firefox 4 and 5, we've learned that there are many more non-hosted Add-ons than we previously thought, mostly those installed by other software. Whether users make use of these Add-ons or not, seeing an incompatible add-on prevents many users from upgrading to new versions of Firefox.
More than 90% of Add-ons are compatible from one version of Firefox to the next, and the ones that aren't usually have binary components that will need to be recompiled every release. If we change the default compatibility assumption, we can reduce the action needed to only the very small number of Add-ons broken by the new release, rather than 100% of add-on developers.
We would move to a model where Firefox assumes Add-ons are compatible unless they have binary components, the author has explicitly said to strictly observe stated compatibility, or it is reported to AMO as incompatible through the Add-on Compatibility Reporter.
It's possible and likely we will occasionally enable add-ons that don't actually work, but as we omit add-ons with binary components, this should not cause crashes and users will simply not have the functionality they expect, and will complain to the add-on author.
This process would not remove the need for AMO's automatic compatibility bumping, as it's still preferred for authors to be aware of changes in Firefox and learn about upcoming deprecations and other changes.
Soliciting feedback at Google groups - mozilla.dev.extensions:
http://groups.google.com/group/mozilla. ... 7c18f037d#
- Note: New Info posted here: