Add WebDriver

[autonome]

• moved files
• removed draft date
• determined name
• added description (triangulated from mdn, spec, caniuse)
• added entry point
• verified baselines
• added to a group, or not
• determined whether the feature should be split
• looked for other related draft features that could be included
• what else?

Resources

• Spec: https://w3c.github.io/webdriver/
• MDN: https://developer.mozilla.org/en-US/docs/Web/WebDriver
• CIU: https://caniuse.com/?search=webdriver
• Developer
  • SO: https://stackoverflow.com/questions/tagged/webdriver
  • NPM: https://www.npmjs.com/search?q=webdriver

Name

No real issues here. There's a "2" in there but developers refer to it as WebDriver everywhere I looked.

Description

Took a bit from both spec and MDN.

Group

Interesting one since it's not really a "web API" in the typical sense. No group for now.

BiDi is already added as feature/webdriver-bidi.yml.

Perhaps we need a WebDriver group, or should merge? Doesn't need to block landing the core of this though.

Splits/Merges

There's BiDi, but it is a "late addition", single vendor [1] and seems to be referred to separately. So not merging at this time.

1. comment below says multiple vendors shipping BiDi

[petele]

Couple of comments

• I'd rename this simply webdriver.yml, drop the "2" since it's not used anywhere else
• Looks like it's missing the VirtualSensor APIs (webdriver.commands.CreateVirtualSensor, webdriver.commands.UpdateVirtualSensorReading)

CC @ddbeck - review the reviewer

LGTM - want @ddbeck to review the reviewer

[autonome]

Thanks for the review @petele!

* I'd rename this simply webdriver.yml, drop the "2" since it's not used anywhere else

✅

* Looks like it's missing the VirtualSensor APIs (`webdriver.commands.CreateVirtualSensor`, `webdriver.commands.UpdateVirtualSensorReading`)

Thanks, added!

[petele]

This is failing npm run test because of the compute_from line. Accepting the suggestion should fix that.

[autonome]

This is failing npm run test because of the compute_from line. Accepting the suggestion should fix that.

hm, worked locally w/ dist, so maybe we should have that script fail on this also

LGTM - want @ddbeck for review of review.

[petele]

hm, worked locally w/ dist, so maybe we should have that script fail on this also

Agreed, it's caught me up a couple of times. I usually will run npm run dist && npm run test in the same line just to double check. Can you file an issue for this please?

[autonome]

flipping back to draft while we work out whether "later additions" are separate features or not per @jgraham's comment on chat:

I haven't looked at the details here, but anything outside the main WebDriver spec can't be considered part of the same feature as the core spec.
Apart from anything else, the implementation status of those extensions is very different

[juliandescottes]

Just a fly-by comment about the summary of this PR:

There's BiDi, but it is a "late addition", single vendor and seems to be referred to separately. So not merging at this time.

BiDi is an extension to the WebDriver specification, but it's not single vendor (assuming that means implemented only in one engine/browser here). It's in implementation in Firefox and Chromium at the moment. See https://wpt.fyi/results/webdriver/tests/bidi?label=experimental&label=master&aligned, Firefox, Chrome and Edge all pass a significant number of BiDi webplatform tests.

[autonome]

Just a fly-by comment about the summary of this PR:

There's BiDi, but it is a "late addition", single vendor and seems to be referred to separately. So not merging at this time.

BiDi is an extension to the WebDriver specification, but it's not single vendor (assuming that means implemented only in one engine/browser here). It's in implementation in Firefox and Chromium at the moment. See https://wpt.fyi/results/webdriver/tests/bidi?label=experimental&label=master&aligned, Firefox, Chrome and Edge all pass a significant number of BiDi webplatform tests.

oh that's great, thanks for the context @juliandescottes!

i'll check that the underlying data here is correct, and will fold it in.

[autonome]

@ddbeck a couple of notes with this one:

Baseline status: Navigator.webdriver is baseline high, but everything under it is all over the place, so I'm leaving as false.

Group vs merge vs neither: Going w/ neither at this point. BiDi is a whole different paradigm, so it's pretty hard to consider it the same "feature" as WebDriver Classic. We could group, but don't need to block on that now - maybe we have a testing or tools group or something in that direction down the road.

[Elchi3]

• Looks like it's missing the VirtualSensor APIs (webdriver.commands.CreateVirtualSensor, webdriver.commands.UpdateVirtualSensorReading)

fwiw, I would put these in a separate feature. These are defined in a separate spec (Sensors) https://w3c.github.io/sensors/#automation.

The same is happening for Compute Pressure, coming to BCD with mdn/browser-compat-data#24211 for which I would have created an own web-feature ID, too.

Basically these things are "WebDriver extensions".

[autonome]

Basically these things are "WebDriver extensions".

ack, per chat:

• broke out sensors commands into separate feature
• added a webdriver group
• added webdriver, bidi and sensors to the new group

@Elchi3 lmk if the webdriver sensors feature looks like what you had in mind

[Elchi3]

Structure looks good to me. Made a suggestion for the description.
r+

[ddbeck]

Everything here looks fine, but I don't think the sensors stuff is ready to ship with the general webdriver stuff. See my line comment for a suggestion. Very close to done though!