<foolip> RRSAgent: make logs public
<gsnedders> ScribeNick: boaz
<astearns> “The name comes from earlier variants on the same theme”
<scribe> scribenick: boaz
presentation beginning
foolip: ill be presenting some
updates on our work on wpt for the last year
... wpt is our collection of tests for most of the web
platform
... in the last year we've added 5k tests (a lot from
css)
... also since last year is more results on more browsers
... worked with bocoup to improve the documentation this last
year. we added search. we deduped and made the content easier
to navigate.
<foolip> slides are https://docs.google.com/presentation/d/1DqzI_GhOE86CKBjJj0SPhf2oUxNE5w94dYzuACGa2oc/edit?usp=sharing
<gsnedders> https://web-platform-tests.org/ for new documentation
foolip: slides are
https://docs.google.com/presentation/d/1DqzI_GhOE86CKBjJj0SPhf2oUxNE5w94dYzuACGa2oc/edit?usp=sharing
... we improved fuzzy reftests
astearns: how many have fuziness turned on
jgraham: a bunch
foolip: added comparison tool for
ref test results
... we also added filtering to the the search bar of
wpt.fyi
... live demo of the filtering and reftest comparison
... priorities for 2019/2020 are to 1) improve interop, looking
at driving down browser specific failures
... ... 2) making PR results more actionable and reliable by
taking our pr bot which runs tests, and making the output tell
you what to do
... 3-7) also test failure triage, search test by metadata,
mobile testing, address flakiness, and wpt.fyi ui improvments
(ease of use, ally)
astearns: there are a ton of open PRs on wpt. where does getting through that bag log fit in your priorities?
jgraham: we have a reviewer system that consists of yml files that name people as reviewers on a per directory basis
zcorpan: if there is a subject matter expert not in there, you should ask them to add themselves
jgraham: we have a fundamental problem with this, because we can't get reviewers for certain technologies
astearns: I think thats fair. I know a lot of the PRs are on CSS tests. so I think it is on my to push people towards adding themselves as reviewers.
<foolip> boaz ^
AmeliaBR: on the issue of pr backlogs, I think it is worth having a discussion about making better tooling to have things like reassigning issues with no reviews after a week, or on a semireg basis sending an email to the working group about their unreviewed tests
astearns: I could try that manually for CSS
<foolip> present lukebjerring
lukebjerring: there is a tool called github-health.appspot.com that uses the google philosophy.
<zcorpan> https://github-health.appspot.com
foolip: there is a tradeoff, sometimes you have to close out reviews that you can't get to, or sometimes you have to go to amillion. crbug is at amillion.
JohnJansen: there is not an easy mapping between wpt dir structure and wgs and specs.
jgraham: it is supposed to be good.
JohnJansen: asks a q to prove that it is not
jgraham: the PRs get labeled
JohnJansen: at my work, I want to go review some tests. I'm not in the wg or working on a spec, but I want to find the technology area that I am interested in
jgraham: im not sure this is
possible
... labels should address this
zcorpan: this is similar to how
to find a test for the feature I'm interested in
... you can use grep to do this
jgraham: good idea to email wgs, if you try that in css, please let us know how it goes.
astearns: experimenting manually will help me figure out what to do
<Zakim> AmeliaBR, you wanted to ask about managing not-ready-to-merge PRs
boaz: will you summarize?
astearns: yes
AmeliaBR: in addition to not getting reviews, and sometimes the review leads to a spec issue.
gsnedders: we have a label for this in the PR
AmeliaBR: can I get docs to that?
<foolip> boaz: let's talk later
scribnic none
foolip: yah, or link that in the 2020 priorities doc
scribenic: boa
... boaz
<JohnJansen> scribenick: boaz
foolip: yes I have an issue open
JohnJansen: it sounds like some members of css wg have filed tests expecting someone outside of the wg to review
astearns: no, its an internal
complaint
... what about the meta data tagging system
lukebjerring: we are thinking of
adding a manifest
... we have a pr for showing open PRs in wpt.fyi
<JohnJansen> +1 to lukebjerring design to let me see PRs open for current directory. that would be much better for my mental model than searching labels
JohnJansen: +1s to lukebjerring design to let me see PRs open for current directory. that would be much better for my mental model than searching labels
foolip: we have identified in the past that test automation is tricky. is that something people are blocked by?
AmeliaBR: im not sure what you mean by automation, but I love the wptests.center
lukebjerring: we're going to integrate that
AmeliaBR: to be able to use that for reftests that'd be great
<JohnJansen> temporary url for feedback on test authoring: https://wptest.center/#/new
foolip: I was refering more to manual tests. how often do you have to write a tests for manual tests
AmeliaBR: we have a legacy issue
with svg, where our existing tests were from a manual testing
era
... manual comes up with interaction tests. there could be a
way to automate those.
foolip: we have a way to automate
mouse/keyboard input, we support that in all the browsers so
long as you are writing testharness.js tests
... do you need it for ref tests?
<BitBot> (14wpt) [PR] jgraham 03merged 13#19029 into 07master: [Gecko Bug 1574690] P3 longer timeout to ease intermittent COEP test - https://git.io/JemhQ
AmeliaBR: we do have declaritive animations that should work with js turned off
foolip: that is tricky
... how many manual svg tests to you have
?
gsnedders: 514 manual tests in svg
boaz: has anyone recently written a manual test?
AmeliaBR: we did one recently because doing it in an automated way was too hard
<BitBot> (14wpt) [PR] jgraham 03merged 13#19116 into 07master: Fixed lint error in #18886 - https://git.io/JeOgS
foolip: tell us when you run into this and we can advise you
zcorpan: also please file an issue with type:untestable
lukebjerring: if we fix the issues that make svg need 514 manual tests, whats the liklihood that they will be replaced?
AmeliaBR: that's a big to do
SimonSapin: in the case of SVG I think its not just conversion work. buttt its hard to make a reference for some svg features without using the svg features
AmeliaBR: this has definitly been an issue
zcorpan: my immediate thought about how to automate those tests is to have an image
<foolip> boaz: let's keep going with this topic for a bit
zcorpan: and then add fuzziness
<astearns> Replacing actual eyeballs with fuzzy matching
lukebjerring: is there tooling that could walk you through manual tests and generate reftests from that
gsnedders: problem with that is text
jgraham: using images as references isn't a great pattern
boaz: how about mdn survey data and wpt data
kadir: now that we know that interop is a big problem it could be nice to dive into what the specific issues are
foolip: work on a compat/interop
follow up survey, then look at what we have in wpt and figure
out what to follow up on
... we want to see these things line up but my guess is that it
wont
AmeliaBR: one thing on this is that the wpt approach to interop is different than the web developer's. because wpt only focuses on latest browser, where as web devs have to support old versions.
rachelandrew_: I agree, that a lot of the issues will be older browsers
boaz: we should have a consistent vocab between wpt and mdn survey
lukebjerring: yes, i agree
... it could be good to name a feature in wpt, and ask a dev if
they think the feature is interoperable
boaz: not sure if we can ask web devs about this in terms of wpt
AmeliaBR: I think we'll need to
group specs together to ask people
... I also think we'll need to separate frustrations from
legacy browsers vs current ones
<BitBot> (14wpt) [PR] Honry 04closed 13#18886: [sensor] Upstream sensor layout tests to wpt - https://git.io/fjjly
<BitBot> (14wpt) [PR] Honry reopened 13#18886: [sensor] Upstream sensor layout tests to wpt - https://git.io/fjjly
This is scribe.perl Revision: 1.154 of Date: 2018/09/25 16:35:56 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) Succeeded: s/JohnJansen: ok im satisfied/ / Succeeded: s/greate/great/ Succeeded: s/type untestable/type:untestable/ Succeeded: s/just conversion/not just conversion/ Present: boaz foolip AmeliaBR myles jsbell cbiesinger JohnRiv Rachel Andrew SimonSapin JohnJansen lukebjerring jgraham Found ScribeNick: boaz Found ScribeNick: boaz Found ScribeNick: boaz Inferring Scribes: boaz WARNING: No "Topic:" lines found. WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth WARNING: No date found! Assuming today. (Hint: Specify the W3C IRC log URL, and the date will be determined from that.) Or specify the date like this: <dbooth> Date: 12 Sep 2002 People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option. WARNING: No "Topic: ..." lines found! Resulting HTML may have an empty (invalid) <ol>...</ol>. Explanation: "Topic: ..." lines are used to indicate the start of new discussion topics or agenda items, such as: <dbooth> Topic: Review of Amy's report WARNING: IRC log location not specified! (You can ignore this warning if you do not want the generated minutes to contain a link to the original IRC log.)[End of scribe.perl diagnostic output]