[minutes] 20110202 Web Performance WG

Web Performance Working Group Teleconference
02 Feb 2011

See also: IRC log<http://www.w3.org/2011/02/02-webperf-irc>

Minutes
http://www.w3.org/2011/02/02-webperf-minutes.html
Attendees
Present
Plh, [Microsoft], +1.650.450.aaaa, +1.415.829.aabb, +1.650.214.aacc, NicJansma, Zhiheng, TonyG, JamesSimonsen, JasonSobel, KarenAnderson
Regrets
Chair
Anderson
Scribe
AndersonQuach
Contents

  *   Topics<http://www.w3.org/2011/02/02-webperf-minutes.html#agenda>
1.     Last Call moving forward to Candidate Recommendation<http://www.w3.org/2011/02/02-webperf-minutes.html#item01>
2.     CORS and Resource Timing and User Timing<http://www.w3.org/2011/02/02-webperf-minutes.html#item02>
3.     Summary<http://www.w3.org/2011/02/02-webperf-minutes.html#item03>

      i.        The Web Performance Working Group members will review the document and provide feedback in preparation to go to Candidate Recommendation. Philippe will touch base with the HTMLWG and WebApps WG to solicit feedback for CR on Navigation Timing.

     ii.        Continue with reviewing test submissions for 2/9/2011 including:

    iii.        Continue investigation to provide authorized cross-domain resource timing information, follow-ups on CORS.

    iv.        Anderson will incorporate User Timing changes as discussed in today's meeting captured in the minutes: getMarks(), and the updates as sent in email.

________________________________

<trackbot> Date: 02 February 2011

list the agenda

scribe+ AndersonQuach
Last Call moving forward to Candidate Recommendation

move to agenda 6

plh: I can ping the HTTP and WebApps WG/WebIDL to review the spec.
... what's the timeline for the first public working draft for Resource Timing and User Timing

AndersonQuach: Let's aim for end of Feb.

move to agenda 1

http://w3c-test.org/webperf/tests/submission/Microsoft/NavigationTiming/test_performance_attributes_exist.htm

AndersonQuach: Approved

http://w3c-test.org/webperf/tests/submission/Microsoft/NavigationTiming/test_readwrite.htm

AndersonQuach: Tests that the object is replacable.

TonyG: One more thought, I wonder if we should check var peformance in addition to window.performance. For webkit the code path is different, not sure if this is an issue in other use agents.

AndersonQuach: that should be easy to add to check var performance = "bar";

http://w3c-test.org/webperf/tests/submission/Microsoft/NavigationTiming/test_unique_performance_objects.htm

AndersonQuach: Approved.

http://w3c-test.org/webperf/tests/submission/Google/NavigationTiming/test_document_open.htm

AndersonQuach: two points of feedback, IE does not have the ability to enum "own" properties for the Performance interface.
... Can be worked around.

TonyG: If this is out of scope then we should have a separate test.

http://www.w3.org/TR/html5/history.html#navigating-across-documents

http://w3c-test.org/webperf/tests/submission/Google/NavigationTiming/test_navigate_within_document.htm#1

TonyG: If i add explicit properties we're good with test_navigate_within_document?

AndersonQuach: IE9 will treat fragment identifiers as non-unique navigations.

http://test.w3.org/webperf/tests/submission/Microsoft/NavigationTiming/test_navigation_type_backforward.htm

AndersonQuach: saw some interesting behavior with iframe.contentWindow.history.back() it appears to navigate the parent window.

move to agenda 3
CORS and Resource Timing and User Timing

AndersonQuach: We last landed at safe defaults, a chunky timeline and then we looked into exploring cross-domain accesss.

Zhiheng: Use html header versus http header.

AndersonQuach: I'll look into Access-Control and feasibility in IE in the near future.

move to agenda 2

NicJansma: We like the streamlined interface in the latest mail.
... Not immediately get the marks sorted by timestamp as they are grouped by key. A developer can sort on their own, analytic tools can do the sorting.

TonyG: was thinking about a more compact JSON implementation

NicJansma: I like how compact the new interface is. Not sure how important it is to sort by timestmap.
... The other concern, with efficiency and notion and understanding with an attribute versus API. With an API developers think about how browser does work.
... If it was a marks attribute, exposing to the user as a DOM object, when requested on demand, the user would more likely or access to it is efficient, e.g. iteration patterns.
... The browser, IE would dynamically have to dynamically construct the marks.

TonyG: That's an interesting point, there could be a hybrid, instead of timing.marks, to move to getMarks();

NicJansma: Yep, and it can easily return the compact JSON packed structure.

TonyG: I can see dynamic patterns, performance.marks.a and grabbing performance.marks.b, it may or may not be expensive, but if we are doing at the time of access, a property may lead to issues.

AndersonQuach: I can update it with the constants, getMarks, mark and clearMarks.

TonyG: We might want to check out Episodes from Steve Souders.
Summary

i. The Web Performance Working Group members will review the document and provide feedback in preparation to go to Candidate Recommendation. Philippe will touch base with the HTMLWG and WebApps WG to solicit feedback for CR on Navigation Timing.

ii. Continue with reviewing test submissions for 2/9/2011 including:

iii. Continue investigation to provide authorized cross-domain resource timing information, follow-ups on CORS.

iv. Anderson will incorporate User Timing changes as discussed in today's meeting captured in the minutes: getMarks(), and the updates as sent in email.