Lightweight default browser for ARM

Registered by Paul Larson on 2009-11-02

Mozilla is a fairly heavy web browser and suffers from less than optimal performance on ARM due to issues such as cache size. Lighter-weight web browsers should be investigated, and a more appropriate one should be picked as a default for ARM based releases.

Blueprint information

Status:
Complete
Approver:
David Mandala
Priority:
High
Drafter:
David Sugar
Direction:
Approved
Assignee:
Alexander Sack
Definition:
Approved
Series goal:
Accepted for lucid
Implementation:
Implemented
Milestone target:
milestone icon lucid-alpha-2
Started by
Alexander Sack on 2009-12-03
Completed by
Alexander Sack on 2010-04-16

Related branches

Sprints

Whiteboard

[asac 01-01-10] working chromium-browser packages for armel karmic (and lucid) are available: https://launchpad.net/~ubuntu-mozilla-security/+archive/nss3.12.3/
[asac 18-12-09] package side of chromium should be fine now. we have gyp, codecs and chromium itself using armv7 flags on demand; runtime issues inprogress
[dyfet 10-12-09] there has been a privacy issue raised with respect to Chromium sending search queries and perhaps other info back to Google along with a personal identifier. Is this something we need to address? Do we need to add/carry a patch to disable this behavior?
[asac 10-12-09] i think this is a none issue. but we should check it. afaik chromium does not really send personal identifiers for search. at least not more than with any other browser.
[asac 02-12-09] updated spec to also reflect actions on firefox to improve experience on ARM; also added more work items on benchmarking and comparing chromium, which probably will consume time too.
[asac 08-01-10] there is no new account skeleton. we will patch the homepage before release.
[asac 08-01-10] firefox with fix for thumb2 is in archive. dave-martin-arm can work on the benchmarking
[asac 08-01-10] gyp - the build system generator for chromium is in the archive.
[asac 13-01-10] postponing items to alpha-3 that need archive changes
[asac 29-01-10] tried to use PGO for further ffox optimizations - no luck - done.
[asac 26-02-10] due to concerns about security maintainability we keep chromium in universe and try to support it to learn how doable that is; hence firefox has to stay the default on arm for lucid - but chromium is in the archive so downstreams can pick that if they are brave ...
[asac 26-02-10] comparison wiki is here: https://wiki.ubuntu.com/Specs/ArmLightweightBrowser/BrowserComparison; feature and memory comparision still missing.

UDS Discussion: https://wiki.ubuntu.com/DiscussionArmLightweightBrowser

Based on UDS discussion, a consensus emerged to move forward with Chromium.

Work items lucid-alpha-2:
Make Firefox use lucid default toolchain flags (thumb2, etc.): DONE
Investigate further firefox optimizations, like: Firefox PGO, different optimization flags etc.: POSTPONE
Prepare Chromium package on ARM: DONE
Resolve Chromium armv7 build issues: DONE
Resolve Chromium armv7 runtime issues: DONE
Get gyp in the archive: DONE
Get Chromium package into the archive: POSTPONE
Fix new acct skeleton for Chromium default: DONE
[dave-martin-arm] Benchmark Firefox: POSTPONE
[dave-martin-arm] Benchmark Chromium: POSTPONE

Work items lucid-alpha-3:
[dave-martin-arm] Benchmark Firefox: DONE
[dave-martin-arm] Benchmark Chromium: DONE
Get Chromium package into the archive: DONE
Investigate further firefox optimizations, like: Firefox PGO, different optimization flags etc.: DONE
draft and publish benchmark: DONE
draft and publish feature comparisons: POSTPONE
Final decision on new default browser: DONE
adjust ARM UNE seed to use Chromium or keep firefox: DONE
validate images to contain these changes: POSTPONE

Work items for ubuntu-10.04-beta-1:
draft and publish feature comparisons: DROPPED

Status:
done: chromium-browser and firefox 3.6 in the archive; first run of performance tests; pending: publish feature performance comparison; sort out chromium security process; decision.

(?)

Work Items

Dependency tree

* Blueprints in grey have been implemented.