Firefox/Features/Expose Add-on Performance: Difference between revisions

m
no edit summary
No edit summary
mNo edit summary
 
(12 intermediate revisions by 7 users not shown)
Line 1: Line 1:
{{FeatureStatus
{{FeatureStatus
|Feature name=Expose add-on performance information in Firefox
|Feature name=Expose add-on impact information at AMO and in Firefox
|Feature stage=Development
|Feature stage=Definition
|Feature status=In progress
|Feature status=In progress
|Feature version=Firefox 8
|Feature health=OK
|Feature health=OK
|Feature status note=Awaiting input from AMO.
}}
}}
{{FeatureTeam
{{FeatureTeam
|Feature product manager=Justin Scott
|Feature product manager=Justin Scott & Asa Dotzler
|Feature feature manager=Justin Scott
|Feature feature manager=Justin Scott
|Feature lead engineer=Hernan Rodriguez Colmeiro, Dave Dash
|Feature lead engineer=Hernan Rodriguez Colmeiro, Dave Dash, Justin Dolske
|Feature privacy lead=Sid Stamm
|Feature qa lead=Henrik Skupin
|Feature qa lead=Henrik Skupin
|Feature ux lead=Jennifer Boriss
|Feature ux lead=Jennifer Boriss
}}
}}
{{FeaturePageBody
{{FeaturePageBody
|Feature overview=Users should be informed when an add-on they have installed will slow down Firefox. This information should be presented in the Add-ons Manager along with a link to learn more.
|Feature overview=Users should be informed when an add-on they are about to install or have installed causes Firefox performance and or memory problems. This information should be presented at AMO and in the Firefox Add-ons Manager.
|Feature users and use cases=We should develop a new Add-ons policy, and acompanying AMO and Firefox technology, which says that we will inform our add-on users and prospective add-on users about specific add-ons that can be proved to have a significant and excessive impact on memory usage and/or performance. I propose we test for start-up performance impact, page load performance impact, start-up memory use impact, and memory leaks.


For more information, please see [http://blog.mozilla.com/tag/performance this post].
The policy should specify that we will notify the authors of add-ons as soon as we discover a problem and it should have a grace period for correcting any excessive memory or performance impact. I propose, as a straw-man, half a release cycle or 3 weeks from notification.
|Feature ux design=* [https://people.mozilla.com/~jboriss/dump/olds_em_mocks_with_performance/as_a_percentage.png Early design mockup]
 
If after the grace period the add-on is not brought within compliance, we should add it to a list of ill-behaving add-ons. That list would be consumed by both AMO and the Firefox client. Both apps would use the list to display warnings to users about the impact of the add-on.
 
If after an additional 3 weeks of the add-on being flagged publicly, (again, a straw man) it is not corrected, we should escalate the content of and the and visibility of the warning message.
 
At AMO, I believe these poorly-performing add-ons should be flagged at the add-on install page. The warning should be firm but friendly -- something like "Mozilla testing has determined that this add-on may cause Firefox performance problems". If, after some period if time from the public warning the add-on is not corrected, the message should be escalated to display in search results and with a text that would more seriously deter users, something like "This add-on is known to cause serious Firefox issues. Use at your own Risk". 
 
In the Firefox client, the flagged add-ons should display warning text in the Add-ons Manager (where informed users, or users following the instructions of a support article or other help, would see it,) and for the escalation period, to an infobar.
 
This would give add-on authors 3 weeks to fix problems before potential users and some of the installed base are notified, and another 3 weeks before most potential and all existing users would see the more dire warning.
|Feature dependencies=Add-on testing capabilities, manual or automated.
 
An AMO API for "warned add-ons" and for "doubly warned add-ons"
 
AMO and Firefox code to access this API, get the lists, and display UI based on the lists.
|Feature ux design=* [https://people.mozilla.com/~jboriss/dump/olds_em_mocks_with_performance/as_a_percentage.png Early design mockup]: shows percentage of runtime used for each addon.
* [https://bugzilla.mozilla.org/attachment.cgi?id=535788]: newer mockup of the first stage
|Feature implementation notes=# <del>File AMO bug to expose performance information in the API</del>
|Feature implementation notes=# <del>File AMO bug to expose performance information in the API</del>
# <del>File Add-ons Manager bug to make use of the new fields and display warnings when appropriate.</del>
# File AMO bug to make use of the new fields and display warnings when appropriate.
# <del>File Firefox Add-ons Manager bug to make use of the new fields and display warnings when appropriate.</del>
# Obtain designs for the Add-ons Manager
# Obtain designs for the Add-ons Manager
# Implement the designs in the Add-ons Manager.
# Implement the designs in the Add-ons Manager.
Line 36: Line 53:
}}
}}
{{FeatureTeamStatus
{{FeatureTeamStatus
|Feature products status=tbd
|Feature products notes=For more information, please see [http://blog.mozilla.com/addons/2011/04/01/improving-add-on-performance/] [http://blog.mozilla.com/addons/2011/04/12/update-on-add-on-performance-testing/] [http://alice.nodelman.net/blog/post/addon-performance-testing-updates-and-future-work/]
|Feature engineering status=tbd
|Feature security status=sec-review-complete
|Feature security status=sec-review-needed
|Feature security health=OK
|Feature security health=Assigned
|Feature security notes=complete: 10.05 [[Security/Reviews/Firefox9/AddOnPerf|Notes]]
|Feature security notes=dveditz
|Feature qa status=waiting
|Feature privacy status=tbd
|Feature localization status=tbd
|Feature accessibility status=tbd
|Feature qa status=tbd
|Feature ux status=tbd
}}
}}
Confirmed users, Bureaucrats and Sysops emeriti
2,974

edits