WebApps F2F Meeting @ TPAC 2013 -- 12 Nov 2013

<ArtB> ScribeNick: ArtB

<scribe> Scribe: Art

Admin, Agenda Setting

AB: this morning we will talk about Web Components

… Dimitri will join us from Mtn View

… and Dominc from Tokyo

… After the morning break we will talk about Testing and Interop

AB: we will start Web Components @ 09:30

<dglazkov> trying...

<Travis> Scribe: Travis

<scribe> Scribenick: Travis

Technical Difficulties...

<ArtB> denis, how do we get the Shenzhen conf room to attach to this new bridge?

Web Components

<ArtB> http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0042.html

ArtB: Please take a minute or two to give us a status update

dglazkov: Things are mostly unchanged.
... custom elements have made pretty good progress--kinks worked out.
... next step is a CR-ready doc.
... the basics are there, everything else is additive.
... Shadow DOM keeps going on--perhaps this year it can get to completion
... custom elements may provide something that is simple and good.
... Shadow DOM has been taking over by ???, who is aligning with DOM4
... also switching it to ReSpec
... I'm focusing on custom elements.

<ArtB> s/taking over by???/taking over by mailto:hayato@google.com/

ArtB: Switching over to ryosuke

<ArtB> http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0418.html -> Ryosuke's email

Ryosuke: Highlights key points from email.

dglazkov: Reviewed the email today, and want to address each point, time might be short.
... sounds like you have the right motivation, but cutting out some pieces that we care about to. We want to build something that works in both cases: same origin and cross-origin.

Ryosuke: we think the current design makes cross-origin use cases much harder.

dglazkov: I think there are ways to do it (via Imports)

<rniwa> Travis: s/Ryosuke/Ryosuke/

dglazkov: I think you're looking for encapsulation around shadow dom
... mozilla researchers came up with an elegant solutions for +1 buttons with trusted/untrusted code.
... was hoping we'd taken everything we wrote down and make something good out of it.

Ryosuke: our concern is with the specs going to last call.

dglazkov: There is no changes to custom elements--these specs are all de-coupled.
... I responded to your proposal, and there is one thing I don't think we should do
... the callback proposal.

<ArtB> http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0436.html -> Dimitri's reply to Ryosuke

Ryosuke: Our concern is more about the HTML Imports; document.register, seems fine-no objections.

dglazkov: This spec needs to cook a lot more.

Ryosuke: let's chat about html imports

Jonas: I have the same concern about how this runs script
... same concern as apple, that imports run scripts in importing docuement. Seems like same mistake with script tags
... it's security issue,
... but bigger problem is bad comm protocol.
... this the impetus for ES6 modules pattern. We should reuse some of these ideas in Html imprts.

dglazkov: You should raise this concern
... please raise the concern on the mailing list
... there there be dragons.
... the polymer folks were starting to work on this, so they'll have more expertise.
... I can set general outcome expectations
... imports in same origin--originally we didn't want to run script at all.
... at first imports was like template only
... then it's changed.
... we a providing a good solution to a tricky problem

Ryosuke: we're open to the idea of accessing global doc in imported doc.
... for same origin of course.
... the imported doc can execute any script by default doesn't play with x-origin case.
... with same origin component, you could use x-origin component with having to completely change the way it behaves.

dglazkov: I don't have a ready solution. Need to outline how to address it.
... with imports, you can add other imports (same or x-origin)
... we have some ideas.
... generally problem with imports...
... version management was an issue for bootstrap people.
... I could do so much more--provide stylesheets for the document via imports.
... etc.
... I could bring in all the scaffolding...
... as long as we provide a solution to these guys [bootstrap] then I'll be happy.

slightlyoff: history context:
... imports work today unifies cross-and-same origin imports.
... they are the same
... just like script include
... and it can do what it likes.
... this is how the web works.
... this is how Alex has a job.
... we've thought about how web component system can create a security barrier.
... layout and script boundaries
... basically need what iframes offer.
... I think we can solve this orthogonally.
... shouldn't block web components.
... if we can decompose the essense of iframe separation, we likely can integrate a solution

<slightlyoff> note that I'm not arguing that iframes are the answer

ryosuke: we thought about using seamless iframe for a solution
... except that with a lot of them...
... there are multiple (20-30 iframes)
... not a scalable solution.
... with separate solutions for same origin and cross-origin
... this is bad, we want one solution for both

hober: to echo jonas
... we are learning lessons doing modules
... (from modules)

slightlyoff: I support the idea/goal
... yes, iframes aren't scalable
... mutual auth wrt global access seems reasonable
... we don't have ES6 modules yet.
... so it seems premature to design a system before this.
... I support the goals, etc., but we should revisit after we have ES6 modules

jonas: html imports are only in one browser

slightlyoff: seems like an alignment with modules, so we should work on the one before the other.

hober: no reason to gate one on the other, they both should proceed together.

slightlyoff: I would want them ordered.

ryusuke: we would like ES6 module + imports to be consistent with each other.
... I don't think we need to fully wait for one to wait on the other.

slightlyoff: Yep. Agree
... TC39 process produces a completely integrated spec draft, after which it takes a little while to get to "rec"

<hober> expanding on that: as we work on both, we find and fix incompatibilities. no one is proposing that the platform should have incompatibles mechanism

slightlyoff: something closer to CR may be approximately early next year.
... I don't want to get too far ahead of this.
... we should consider imports with modules together.
... let's avoid crazy vortexes.

dglazkov: +1 on imports + module consistency

hober: violent agreement.

ryosuke: our big concern was if html imports went forward today, we have big concerns with x-origin use cases.
... if things slow down, then we can make sure this is done right.

dglazkov: We need to figure out what those x-origin issues are, merge them, and move forward.
... ryosuke: please drive this!

jonas: I think module discussion and x-origin are unrelated

<slightlyoff> so the other analogs are workers

<slightlyoff> and we don't have workers that have renderers

jonas: want to figure out the x-origin story. Seems hard.
... whatever we create may seem a lot like iframes, or use a lot of the same primitives

<slightlyoff> so if we had a RenderWorker that had the ability to have some sort of DOM view to it, that'd be workable, but again, we don't have that infrastructure

jonas: dosen't seem like a blocker (to me).

dglazkov: Want to write down this proposal.
... sync with web app sec.
... should start looking at x-origin use cases.

ryosuke: new topic (slighty)--templates
... chatted with rafael, etc.
... want to do mustache syntax in templates.

dglazkov: Seems just outside my scope.... may need to work with rafael.

hober: You're comfortable addressing mustache syntax in templates?

dglazkov: Yes, YES, YES!!!
... very comfortable.
... I suspect that we'll hear that there isn't one way to do this
... (it'll be hard to land on "the right one")

hober: not married to the double {{

ryosuki: that's all

Chaals: no further questions..

dglazkov: Keep up the good work

<rniwa> dglazkov: thanks!

Travis: Microsoft hasn't looked in depth at this yet.

<aboyet> scribenick: aboyet

<ArtB> Scribe+ Adam

Test Interopt

artb: wanted to go through specs and what it's going to take to get to next steps
... first one IndexDB

… last college about 6 months ago, Alex agreed test facilitor for that spec

… have ~7 requests for IDB and not seemed to be reviewed yet, anybody give status of this?

adrianba: believe the work alex has been doing is moved away from IDB, we'd be happy with help on it

<ArtB> ACTION: barstow find another Test Facilitator for IDB spec [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action01]

<trackbot> Created ACTION-704 - Find another test facilitator for idb spec [on Arthur Barstow - due 2013-11-19].

artb: will record action to find test person

…. can anybody here to volunteer to help with the IDB testing effort ?

… curious what Mozilla's level of interest is for this

… need to review PRs for IDB as well

<ArtB> http://lists.w3.org/Archives/Public/public-webapps-testsuite/2013Oct/0013.html -> IDB Pull Requests for tests

… 4 diff pull request during test web forward paris event

…. first step for interop is to try to agree on the set of tests, until we get review then we don't have a basis

… really pleading to get people to step for review

james: think there are some overlap in tests from microsoft tests and opera tests

…. other submissions are smaller, so not as much as it looks like there

artb: thanks for clarification …. we need somebody as test facilitator for IDB, won't be able to move forward in interop until tests are reviews

… anything else (to the crowd)

Server Sent Events

artb: not too far off from having 2 impls passing the tests, tina has done some good works in analyzing the failures

… so there are 5 test failures preventing from going to recommendation. from chrome and firefox

james: it looks like there are some common failures … maybe it could be a change to the spec then to fix both of them but depends if failing for same reason so some sluething needs to occur, OK scratch all of that

… it's an interaction with a second spec

chaals: maybe small enough bugs or known bug from before

… just need to look in to the bugs an tina could provide more

<ArtB> ACTION: talk to Tina about how to make progress on SSE interop [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action02]

<trackbot> Error finding 'talk'. You can review and register nicknames at <http://www.w3.org/2008/webapps/track/users>.

artb: what about the timeout one failing is actually a test case issue instead … will talk to tina about it

james: have a feeling these were opera test originally … zzzz is around here could ask him

artb: anything else on event source?

Web Messaging

artb: have alex as test facilitator , adrian is alex able to help with this one ?

adrian: talked to chris, his opinion is we have a good enough set of test for web messaging an there is interop here.

… while still some failures looks like there might be some need investigation

… propose give me action to talk to chris

<ArtB> ACTION: bateman followup with KrisK re WebMessaging test results http://www.w3.org/wiki/Webapps/Interop/WebMessaging [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action03]

<trackbot> Created ACTION-705 - Followup with krisk re webmessaging test results http://www.w3.org/wiki/webapps/interop/webmessaging [on Adrian Bateman - due 2013-11-19].

artb: anymore ?
... ok next up

Web Sockets

artb: Kris is test faciliator i went ahead and ran these, some timeouts towards the end of the tests maybe a problem with the tests themselves

james: yeah, that looks wrong, the web sockets test have historically had issues there

… suggest re-running tests to improve things

adrianba: may have some issues with network configs etc. proxy server

… conclusion we had they were until there were changes to people impls there were issues that still needed to be addressed

… still a question about how importatn are those test in grand scheme

artb: what is reasonable next step?

… does it make sense to have Kris eval what is there and make a recommendation?

adrainba: don't think they are bogus, the opera tests added added additional coverage

… one option would be to analyze the current coverage of the tests and what the failures are a resulting from

james: think those results are misleading, just ran the tests on gecko and we are def. getting some passes where those results are getting timeouts

… just for anybody looking at the screen and seeing red

artb: would you be willing to update the gecko results?

james: yeah but there is a problem with running the secure tests so get timeout for anything that runs the ssl tests but can update for non-ssl tests

artb: leaves somebody to run the chrome ones again, anybody from opera

richt: yeah we could run them again

<ArtB> ACTION: pieters run the Web Sockets tests and put results in http://www.w3.org/wiki/Webapps/Interop/WebSockets [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action04]

<trackbot> Created ACTION-706 - Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets [on Simon Pieters - due 2013-11-19].

<ArtB> ACTION: bateman Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action05]

<trackbot> Created ACTION-707 - Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets [on Adrian Bateman - due 2013-11-19].

<ArtB> ACTION graham Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets

<trackbot> Created ACTION-708 - Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets [on James Graham - due 2013-11-19].

artb: anybody willing to run chrome tests?

<ArtB> ACTION Run the web sockets tests and put results in http://www.w3.org/wiki/webapps/interop/websockets for Chrome

<trackbot> Error finding 'Run'. You can review and register nicknames at <http://www.w3.org/2008/webapps/track/users>.

richt: i'll talk to simonp about running chromium tests

Web Workers

artb: have an interop table but not tests have been run it seems

… does any implementation implement ?

… if the answer is no then probably not worthwhile to run tests

james: i know gecko doesn't ….. i don't know about others

alex: not authoritative on this topic ….there are issues with some stuff but not aware of all of them

chaals: but does sound like worth running tests suite

… on chrome

artb: any volunteers to run it ?

kinuko: I can run them

<ArtB> ACTION: kinuko run the Workers tests on Chrome and put results in http://www.w3.org/wiki/Webapps/Interop/WebWorkers [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action06]

<trackbot> Error finding 'kinuko'. You can review and register nicknames at <http://www.w3.org/2008/webapps/track/users>.

adrianba: feeling like ground hogs day on this

… it seems like almost every tpac we have this same conversation

artb: a tpac or two ago I proposed the same thing and got bashed for it; , would need somebody to step up to do the editorial work

chaals: anybody think that would be a fun thing to do and use the delete key all over the spec

travis: volunteers to take that up

<ArtB> ACTION: travis remove the Shared Workers from the Workers spec [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action07]

<trackbot> Created ACTION-709 - Remove the shared workers from the workers spec [on Travis Leithead - due 2013-11-19].

artb: it still makes sense to run the tests then

… thats all i had for interop and testing

chaals: gives travis some crazy bottle

Testing

<ArtB> AB: thanks James for leading this session!

<ArtB> Scribe+ ArtB

<ArtB> ScribeNick: ArtB

JG: this afternoon is mainly about testing

… it's important for a number of reasons

… needed for advancement to REC

… but more importantly, we need tests to get the interop we want

… deviations in impls means web devs hate us

… and then they write native apps instead

… Although we have in WebApps some tests, we are having problems getting some tests reviewed

… and without review, the tests are not "accpeted"

… So, I'll take about Test Review this afternoon

[ James shows slide regarding Reasons for Test Review ]

JG: 1. Check tests are correct per spec

… 2. Check tests are written robustly

… 3. Check for style issues that will make tests fragile or hard to run

… 4. Ensure tests are not using server-side script in a malicious way

JG: don't want someone from outside of W3C doing bad stuff like %rm -rf /

… I'll start with some tools for test review

… Re test review @ w3c

… Start by cloning web-platform-tests (on GH)

… We have many Pull Requests (PR) now

… The simplest tool is to use GH directly

… and just add a comment to a commit

… GH can show raw diff

… Just click on a line number and then add a comment

… (Here's an example PR: https://github.com/w3c/web-platform-tests/pull/63)

… It's hard to know if a comment has been addressed

… Also hard to know if a review is complete or not

… Although GH for code review is ok, it isn't always power enough, especially for more complicated reviews

JR: we have the "Critic" tool or Opera Critic for reviews

… https://critic.hoppipolla.co.uk/home

… Opera created this because other code review tools were not sufficient for some uses

… it can be a bit intimidating at first

… Critic use GH credentials (via OAuth)

… It will try to complete a profile for you

… At a minimum, need to fill in your email to get notifications

… For reviewers, can configure it so that it only shows reviews for those reviews of interest

… (unlike GH which has a very coarse notification system)

… Go to Filters and then Add filter to set selections

… < James shows a Critic review >

… Even if you haven't created a filter, you can still submit a comment

… One main advantage of Critic is that it says, on a per file and per commit basis, the percentage of the review that is complete

… it also tracks the review issues that are still open

… One question then is, how do I do the review?

… Can select the commit to review

… If a review has a bunch of commits, can select all of the commits and then see all of the proposed changes for the entire review

JG: If issues are raised, they must be closed before a PR can be merged/closed

… when submitting a comment, can differentiate "Issues" from "Notes"

… It is also possible with Critic to make an Issue span more than one line of code

JG: if someone wants to use Critic with a PR, please let me

… I think it has significant advantages over GH's review feature

RN: is critic feedback sent to GH?

JG: no, think that would result in lots of e-mail

RN: it would be nice if GH and Critic could sync

JG: that's possible but probably not trivial

… think it would result in an annoying amount of emails

… In practice, that hasn't been a problem so far

JG: those are the two tools for code review

… The next Q is, "what should I be looking for?"

… the TestTWF site has all of the appropriate docs

<aboyet> http://testthewebforward.org/

JG: < Displays Slide for Test Review Checklist >

… Use most specific assert possible

<rhauck2> http://testthewebforward.org/docs/

… Make it clear if there is an "Expected Failure"

… Dont use internal timeouts unless they are unavoidable. Prefer events

… Use idlharness.js where possible to check WebIDL-derived assertions

… External resources must not be used

<rhauck2> this checklist should be integrated here: http://testthewebforward.org/docs/review-checklist.html

… Check files don't have the executable bit set if not necessary

… Ensure server-side code does not do anything malicious

<scribe> ACTION: graham integrate Test Checklist with http://testthewebforward.org/docs/review-checklist.html [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action08]

<trackbot> Created ACTION-710 - Integrate test checklist with http://testthewebforward.org/docs/review-checklist.html [on James Graham - due 2013-11-19].

… I've recently submitted a PR to remove PHP

<rniwa> rhauck2: do you know what's happening there?

JG: < Code Style slide >

… Avoid trailing whitespace

… Use sufficient whitespace to make sure that tests are differentiable

… Prefer spaces to tabs

… Use Unix-style LF instead of Windows-style CR LF

<rhauck2> rniwa: oh dear - we had a great firewall problem with the site the other day in a different room

RN: can't these checks be automated?

<rhauck2> rniwa: maybe we're hitting it now here?

JG: yes, certainly in theory

<rniwa> rhauck2: i'm on VPN though

<rhauck2> rniwa: i'm currently vpn'ed in

<rhauck2> rniwa: me too

<rhauck2> rniwa: does staging work for you? http://www.testthewebforward-staging.org/docs/

RN: perhaps Critic could automatically check some of these style issues

JG: yes, it does for some stuff (like trailing WS)

<rniwa> rhauck2: yup

<rniwa> rhauck2: http://www.testthewebforward-staging.org/docs/review-checklist.html works too :D

<rniwa> rhauck2: thanks!

<rhauck2> ArtB: btw, cross logged the AI on the testtwf-website issue tracker: https://github.com/w3c/testtwf-website/issues/194

<rniwa> rhauck2: is this some sort of a server issue?

<rhauck2> rniwa: weird. whatup with Apple?

<rhauck2> rniwa: no idea

JG: it is possible to run all PRs directly

<rniwa> rhauck2: oh, it's us :/ nm...

<rhauck2> rniwa: works for me both on and off vpn

<rniwa> rhauck2: but i can access http://testthewebforward.org

JG: f.ex. http://w3c-test.org/web-platform-tests/master/workers/

<rniwa> rhauck2: only http://testthewebforward.org/docs/ is what i get 404 on

<rhauck2> rniwa: (we fixed the great firewall problem when we discovered it the other day - had to do with a godaddy name server that was blocked in china)

… this automated mirroring can cause probs f.ex. if a PR included bad php

<rhauck2> rniwa: what about stuff under /docs? like: http://testthewebforward.org/docs/testharness-tutorial.html

… To get something mirrored to w3c-test.org/web-platform-tests/

… if one isn't in the "special grou

<rniwa> rhauck2: that one worked

<rniwa> rhauck2: just http://testthewebforward.org/docs/review-checklist.html and http://testthewebforward.org/docs/ I guess

<rhauck2> rniwa: weirdo man

… group, then someone in that group needs to add a w3c-test:mirror comment to get the mirror to happen

<rhauck2> rniwa: just the exact 2 pages we were talking about :)

<rniwa> rhauck2: yeah :(

JG: so that's all I had re the mechanics of Code Reviews

… there are now 127 Open PRs!

… thus a backlog

RB: we got about 30 PRs during the TestTWF Shenzhen event on Nov 9

<rniwa> rhauck2: huh, it started working now. so strange. thanks for the help

AB: can we get a list of PRs for WebApps?

RB: hmmmm

JG: the answer is yes, but no URL

<scribe> ACTION: barstow sent James a list of WebApps' test dirs in w-p-t [recorded in http://www.w3.org/2013/11/12-webapps-minutes.html#action09]

<trackbot> Created ACTION-711 - Sent james a list of webapps' test dirs in w-p-t [on Arthur Barstow - due 2013-11-19].

ArtB: we need review help with several PRs

… especially IDB (we have 7 open PRs)

… If anyone can commit to a review, please let me know

JG: PR 379 is an important one too

… if anyone can look at that, please let me know

… platform

… https://github.com/w3c/web-platform-tests/pull/379

JG: re PR379

… need review some of the XHR tests

… also NavigationTiming test need review

PLH: I already submitted a PR for the NavTiming part of 379

GH: need review of async tests

… some Opera dnd tests

… script scheduling tests (from O)

… preload tests (expect Simon to review those)

ArtB: here are some open PRs for WebApps

… File API: https://github.com/w3c/web-platform-tests/pull/53

… IDB: https://github.com/w3c/web-platform-tests/pull/63

… IDB: https://github.com/w3c/web-platform-tests/pull/634

… Shadow DOM: https://github.com/w3c/web-platform-tests/pull/143

… HTML Imports: https://github.com/w3c/web-platform-tests/pull/178

… Shadow DOM: https://github.com/w3c/web-platform-tests/pull/194

… Shadow DOM: https://github.com/w3c/web-platform-tests/pull/278

… File API: github

… File API: https://github.com/w3c/web-platform-tests/pull/279

… File API: https://github.com/w3c/web-platform-tests/pull/282

… File API: https://github.com/w3c/web-platform-tests/pull/283

… File API: https://github.com/w3c/web-platform-tests/pull/284

… EventSource: https://github.com/w3c/web-platform-tests/pull/285

… File API: https://github.com/w3c/web-platform-tests/pull/286

… File API: https://github.com/w3c/web-platform-tests/pull/2867

… File API: https://github.com/w3c/web-platform-tests/pull/286287

AB: and more PRs for Web Apps …

… File API: https://github.com/w3c/web-platform-tests/pull/287

… File API: https://github.com/w3c/web-platform-tests/pull/288

… File API: https://github.com/w3c/web-platform-tests/pull/289

… File API: https://github.com/w3c/web-platform-tests/pull/290

… File API: https://github.com/w3c/web-platform-tests/pull/291

… File API: https://github.com/w3c/web-platform-tests/pull/293

… Progress Events: https://github.com/w3c/web-platform-tests/pull/294

Writing Tests

JG: I'm not sure what we want to try to accomplish in this session

… We are getting more contributinos

… If anyone has specific questions about "how to test X", I can work with them 1:1

<aboyet> http://testthewebforward.org/docs/

… A new exciting resource is testthewebforward.org

… Having public events

… to write tests

… to improve interop

… It has become the "overall" name for W3C Testing Effort

… which is great [to have that brand]

… The site has lots of docs for tasks about writing tests

… Rebecca is the expert

… and she is hosting a session about the Future of TestTWF tomorrow

… Our test repo is GitHub (GH)

… Need to start with git and how to submit tests

… The TestTWF tutorial is great

… And one can always ask Q's in W3C's #testing or #webapps IRC channels

… As of today, there are 3 types of tests

… 1. JavaScript

… uses testharness.js

… I can help anyone with the harness

… 2. RefTests - for testing rendering

… relevant f.ex. for CSS

… could be needed by WebApps, f.ex. if the test has some "display"

… Instead of just creating one test, need a second file

… that serves as the "reference"

… This allows Pass/Fail based on comparing both files

… 3. Manual Tests

… we try to avoid these because they are costly

… f.ex. time consuming to run

… Only want those if #1 or #2 can't be used

… Also in the future expect to accept Web Driver tests

… allows driving the browser externally

… < James introduces David Burns (DB) >

DB: I co-edit Web Driver spec

… allows automated driving the browser simulating a user

… we aren't "stuck" in JavaScript sandbox

… We can do trusted events (e.g. logging into a specific service)

DB: < gives a demo … >

<aboyet> http://www.w3.org/TR/webdriver/

DB: Web Driver starts the browser, sets default config

… we can then do things like navigate to URLs

… can use Python, C#, Ruby and a couple of others

… FB has PHP version

… anyone can write a language binding (spec says how to do so)

… can execute arbitrary JS

… the JS is sandboxed

DB: we also have advanced user interactions

… f.ex. shift+letter+mouse_drag can be done

… really nice for doing things outside of a JS test framework

AB: nice!

DB: implemented in FFox, Chome and Opera

… we use it to test FFOS

… We also allow it to access browser chrome

… If testharness.js doesn't fit, please try Web Driver

… There is interest from other browser vendors

Travis: there is a 3rd party soln for IE

DB: we hope all browser vendors will implement it

… Please let me know if you want to chat about Web Driver or get some demos

JG: one change that is coming soon for WPT is a self-contained Python environment

… and that means dropping PHP

… think this will make some tests easier to write

… This eliminates installing some 3rd party stuff

… Now have to clone the repo, install several packages, edit a host file

… But with the changes, just run one Python script

… and then access the tests via a browser

… (f.ex. web-platform-test:8000/websockets/)

… So can run the test suite locally

Travis: Sweet!

JG: script is on GH

… as well as the README file

… If you need a complex test (e.g. multiple domains or ports), might need to read server-side.md

… There is also a "Pipes" feature

… they can be used to do things like changing headers

… Useful for XHR testing (e.g. if need to wait for the file to get loaded)

… Instead of hardcoding domains or ports, use the templates syntax

… and the replacement will occur (i.e. template resolved) when the tests are run

JG: we have "asis" file support

… and can also script the server via .py files

… < see "py files" in the documentation >

JG: the documentation includes some examples

… https://github.com/w3c/web-platform-tests/blob/jgraham/python_review_2/server-side.md

< Applause for James! >

- DRAFT -

WebApps F2F Meeting @ TPAC 2013

12 Nov 2013

Attendees

Contents