[minutes] Web Performance WG Teleconference #139 Agenda 2014-10-15

Highlights:
* Resource Hints merged pre-render and pre-load
* Web Smoothness meets no major technical objections from IE team
* Page Visibility 2 should be discussed and not dropped yet
* Error Logging should likely be more specific about failed attempts and include batching in the spec
* Navigation Timing should expose protocol information



Resource Hints
- Merging Pre-load and Pre-render based on feedback
- Will plan to move this to first public draft with Philippe's help

Web Smoothness 
- IE already exposes this information in ETW traces. 
   Microsoft doesn't have any strong objections to exposing it in real-time through JavaScript.
- Some language could be tweaked: 
   Would prefer not to use vendor-specific lingo such as "jank" in the spec
   Naming of the spec should better match what it is to prevent mischaracterization (aka, it doesn't make sites "smooth")
- Move over to Github with Philippe's help
- Move over conversation from Mozilla as well

Beacon 
- No feedback / discussion on the call

TPAC
- Page Visibility 2 - should be in agenda
   Are there objections to it?
    - Not Microsoft
    - Chrome might have had some objections to it. Arvind and Ilya will investigate.

Error Logging
- Should we retry if it fails to upload?
   Can we defer to the UA implementation?
   -   Not a great idea because it causes implementation issues
   -   Not protocol data - attempt to deliver it

- Batching
   Requires a change to the definitions
   Can de-dupe information
   Should consider batching

- Individual resources on the page
   Could include resources that failed to load in Resource Timing 
   Truly about the HTML page navigation

- Privacy
   Should we scope error reports to origin.
- Start time
   Is there an ability to delay the report
   Google security time has had a lot of spurious alerts firing because clients have incorrect clocks
- Question the utility of start time UTC
   If we do delay it, should provide how long it has be delayed for
   Can get tricky in batched reports. Single header wouldn't work.
   - Should discuss issues on the group 

Navigation Timing
- Exposing Protocol
   Something we have discussed for a while/in the past
   Based on feedback, we might have proposed text
    Should it apply to resource and navigation timing
    Should seek input from all browsers
   IE will consider. Don't see reasons to object currently as it's exposed in dev tools and ETW already.
-  Transfer and decoded size
    Better language was suggested and it's in the github issue
- Follow up on the list to see if anyone has feedback.


Tobin

Received on Wednesday, 15 October 2014 19:59:02 UTC