Custom handlers automation #8267

javifernandez · 2022-09-06T08:40:07Z

This commit adds a new subsection about User Agent Automation in the Custom Handlers section. The purpose of this new subsection is to define a new WebDriver extension command to define auto-replies for the prompt dialog that the spec suggests to offer to the user in order to grant the registration of a new protocol handler.

Fixes #8251.

At least two implementers are interested (and none opposed):
- User Agent automation for Custom Handlers mozilla/standards-positions#685
- https://groups.google.com/a/chromium.org/g/blink-dev/c/VVW1t8xeXXc/m/YImce8AwCAAJ
Tests are written and can be reviewed and commented upon at:
- Test automation for custom handlers web-platform-tests/wpt#35792
Implementation bugs are filed:
- Chrome; https://bugs.chromium.org/p/chromium/issues/detail?id=1359103
- Firefox: https://bugzilla.mozilla.org/show_bug.cgi?id=1788789
- Safari:

/index.html ( diff )
/infrastructure.html ( diff )
/references.html ( diff )
/system-state.html ( diff )

source

domenic · 2022-09-07T03:43:11Z

source

+
+  <hr>
+
+  <p>The <dfn>current registration automation mode</dfn> tracks what


What does the "current registration automation mode" belong to? Is it user-agent wide? Window? Browsing context?

My gut feeling is that it should belong to a Browser Context, but I'd appreciate comments and / or suggestions on that regard.

Since this flag/attribute is intended for testing purposes only, and give than we might encourage that the user agent first verify that its handled in an automation context, it makes sense that its indeed associated to a browser context.

My default suggestion would be Document. I would be surprised if you set this, and then navigated to a totally different page (e.g. on another origin), and the setting persisted.

However, if such browsing context-wide impacts are normal for WebDriver, then maybe browsing context is better.

I admit you are right that is difficult to imagine a scenario where is useful to keep this flag across different Documents. However, many of the currently defined WebDriver commands "happen in the context of either the current browsing context or current top-level browsing context".

Additionally, if I've understood it correctly, the WPT's testdriver,js, which is the JS implementation of the WebDriver client, is expected to operate across different browser context.

I'm not sure whether restricting the use of the "current registration automation mode" to the Document would limit how we could use it to define certain tests.

The Secure Payment Confirmation spec, which inspired me to define this new extension command, doesn't specify where the current-transaction-automation-mode belongs to. I agree we could do better in this PR and specify clearly that this flag is stored either in the current browser context or in its associated active document, whatever we end up agreeing on.

I'd like to get feedback from @jgraham on this regard, but my opinion is still that using the Browser Context is more convenient in this case.

Most current WebDriver commands aren't setting state that persists across navigation, so although they operate at the level of browsing contexts, the action of the command is on the active document in that context.

The exceptions are mostly WebDriver-internal state which is mostly per session (i.e. global). Argably the closest analogue to this feature is the User Prompt handler, which is a global setting which affects what happens when an alert appears. I don't think that's good design, but in the context of WebDriver it makes sense; in particular if the handler was per-document then it wouldn't be possible to reliably setup the handling before an alert appears.

So I think the question is: do you need to handle cases where registerProtocolHandler is called without user interaction? If not per-document state is probably fine. If you do then it's hard to see how to make that work without the state being global.

At least Chrome, and I believe is Firefox too, require that calls to registerProtocolHandler are done under a transient user activation. Hence, in order to define automated tests we would need to use the test_driver.bless function. You can see some WPT examples in the PR I have for this proposal.

The state we are adding here would only apply under a testing context and only modified via the WebDriver extension command. I agree that its goal is almost the same than the User Prompt Handler; as a matter of fact II tried to implement the automated tests using it, but it only works with simple dialogs so its not an option in this case.

As I commented in the issue where I introduced the proposal, we already have a WebDriver extension command that provides a similar feature, the Set SPC Transaction Mode command. Unfortunately, in that spec it doesn't specify where the state belongs to.

Most current WebDriver commands aren't setting state that persists across navigation, so although they operate at the level of browsing contexts, the action of the command is on the active document in that context.

Based on this comment from @jgraham and @domenic's preference for Document as the owner of this new flag, I'll change the PR to sate it explicitly.

source

domenic · 2022-09-07T03:49:51Z

source

+      argument</span>.</p>
+    </li>
+    <li>
+      <p>Set the <span>current registration automation mode</span> to <i>mode</i>.</p>


Same problem here with what object the current registration automation mode belongs to.

As I said before, I believe this should belong to the browser context.

source

javifernandez · 2022-09-12T14:07:46Z

Thanks for the review. See my comments inline.

javifernandez · 2022-09-20T10:36:34Z

Hi @domenic I pushed some changes, trying to address all the changes requested in the last review. Could you please take another look ?

source

domenic · 2022-09-27T07:49:55Z

source

+
+  <hr>
+
+  <p>The <dfn>current registration automation mode</dfn> tracks what


My default suggestion would be Document. I would be surprised if you set this, and then navigated to a totally different page (e.g. on another origin), and the setting persisted.

However, if such browsing context-wide impacts are normal for WebDriver, then maybe browsing context is better.

source

…ndler step 2.

domenic

OK, mostly editorial stuff, plus the big question of where the current registration mode should be stored, which is still outstanding.

source

…ent.

source

domenic

OK. I made a variety of changes in my last few commits to fix remaining errors. Please take a look and let me know if they look reasonable. In that case, we can proceed to merge this.

javifernandez · 2022-11-22T09:20:01Z

Hi @domenic, thanks for the reviews and the last edits; they all look good to me.

domenic · 2022-11-23T05:46:29Z

Hmm, I realized this doesn't actually have multi-implementer support? I guess you are implementing for Chromium, but you linked to mozilla/standards-positions#685 for Mozilla, which has no replies.

Let me know when you have a second implementer interested, and then we can merge.

javifernandez · 2022-11-23T23:18:17Z

Let me know when you have a second implementer interested, and then we can merge.

Yes, I''ve got some informal support from @jgraham during a meeting to review the WPT RFC , but I agree we should wait until we got explicit support in the position request. There is also an ongoing conversation in the Firefox bug I've filed, but again still not clear support there either.

javifernandez · 2022-12-07T00:55:20Z

Hmm, I realized this doesn't actually have multi-implementer support? I guess you are implementing for Chromium, but you linked to mozilla/standards-positions#685 for Mozilla, which has no replies.

Let me know when you have a second implementer interested, and then we can merge.

There was some recent feedback on the mozilla standard-position request; do you think it'd be enough to move forward ?

domenic · 2022-12-07T03:43:24Z

Let's ask: @jgraham, based on your comments there, would you be happy to see this in the HTML Standard?

javifernandez · 2023-01-10T17:57:20Z

Let's ask: @jgraham, based on your comments there, would you be happy to see this in the HTML Standard?

FWIW, I'm in conversations with some Firefox devs regarding the implementation of this webdriver extension command in Marionette.

jgraham · 2023-01-10T18:21:01Z

I'd say if other implementors are happy with this approach then Mozilla are too.

The PR#8267 [1] for the HTML spec introduced a new WebDriver extension command to allow testing automation for the registerProtocolHandler method of the Custom Handlers API. This CL adds support in ChromeDriver for such extension command. See also the Web Platform Test PR [2] and RFC [3] for further details. The extension command relies on a new CDP command implemented in a different CL [4]. [1] whatwg/html#8267 [2] web-platform-tests/wpt#35792 [3] web-platform-tests/rfcs#121 [4] https://crrev.com/c/4203529 Bug: 1359103 Change-Id: Ifd4385815ab30a40f46c2f083dc9604024060613 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4200660 Reviewed-by: Vladimir Nechaev <nechaev@chromium.org> Commit-Queue: Javier Fernandez <jfernandez@igalia.com> Cr-Commit-Position: refs/heads/main@{#1100968}

javifernandez added 2 commits August 31, 2022 13:52

New section in the Custom Handlers chapter about User Agent Automation

2435977

Fixed some references and typos

5f85e95

javifernandez mentioned this pull request Sep 6, 2022

User Agent automation for Custom Handlers mozilla/standards-positions#685

Open

domenic requested changes Sep 7, 2022

View reviewed changes

javifernandez added 10 commits September 9, 2022 11:12

Moved the new section and added an id for its title

66b8355

Added a new section in References and adapt the dfn elements.

aaf4b26

REmoved h7 heading and joined two paragraphs

510a298

Removed uselss hr elements

0c25779

Change the enum defintion and some related references.

52bb3af

Removed unneeded 'figure' element.

4c18931

Fixed extension-commands external reference.

4200632

Removed unnedded backtricks

1302e1d

Fixed format according o the style rules.

d7cfa05

null is not a variable or italicized

e88380c

javifernandez added 2 commits September 20, 2022 00:40

Add a new secttion about the prompt UX

67c3d82

Replace the IDL definition by a regular paragraph.

81d0761

domenic reviewed Sep 27, 2022

View reviewed changes

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Outdated Show resolved Hide resolved

source Outdated Show resolved Hide resolved

javifernandez added 3 commits October 11, 2022 15:53

Replace the new UX section and instead, modify the registeRProtocolHa…

2aa0598

…ndler step 2.

Following the W3C casing rules for enumerations

63fe161

Several formatting changes

c8a9160

domenic reviewed Oct 12, 2022

View reviewed changes

source Show resolved Hide resolved

source Outdated Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Outdated Show resolved Hide resolved

javifernandez added 6 commits October 19, 2022 00:44

some editorial changes

9d7d262

Use 'code' to properly define these terms as Strings

36eb7fa

Fix <li> identation

e1a949a

Use 'code' instead

e2b1547

Rephrasing as suggested

9400612

Stating clearly that the registration-mode state belongs to the Docum…

38a6b1a

…ent.

domenic reviewed Oct 23, 2022

View reviewed changes

source Outdated Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

foolip mentioned this pull request Nov 1, 2022

RFC 121: Add testdriver.js support for Custom Handlers registration web-platform-tests/rfcs#121

Merged

javifernandez and others added 6 commits November 8, 2022 23:52

Some formatting hcnages, suggested in the last review

6a1c67a

Simplify defs and refs.

df92ee0

Some wrapping fixes

05aa141

Use the current browsing context's active document concept.

a2d9446

Various fixes

860d08b

Extra newline

94f3365

domenic approved these changes Nov 18, 2022

View reviewed changes

jgraham mentioned this pull request Dec 6, 2022

Add consume user activation of Window w3c/webdriver#1695

Closed

domenic merged commit 3ea8e08 into whatwg:main Jan 11, 2023

This was referenced Feb 2, 2023

Custom Handler integration with the Permission API #7920

Closed

Add test API for protocol handlers web-platform-tests/wpt#26819

Closed

javifernandez mentioned this pull request Feb 16, 2023

Test automation for custom handlers web-platform-tests/wpt#35792

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Custom handlers automation #8267

Custom handlers automation #8267

javifernandez commented Sep 6, 2022 •

edited by pr-preview bot

Loading

domenic Sep 7, 2022

javifernandez Sep 12, 2022

domenic Sep 27, 2022

javifernandez Oct 18, 2022

jgraham Oct 20, 2022

javifernandez Oct 20, 2022

javifernandez Oct 21, 2022

domenic Sep 7, 2022

javifernandez Sep 20, 2022

javifernandez commented Sep 12, 2022

javifernandez commented Sep 20, 2022

domenic Sep 27, 2022

domenic left a comment •

edited

Loading

domenic left a comment

javifernandez commented Nov 22, 2022

domenic commented Nov 23, 2022

javifernandez commented Nov 23, 2022

javifernandez commented Dec 7, 2022

domenic commented Dec 7, 2022

javifernandez commented Jan 10, 2023

jgraham commented Jan 10, 2023


		<hr>

		<p>The <dfn>current registration automation mode</dfn> tracks what

Custom handlers automation #8267

Custom handlers automation #8267

Conversation

javifernandez commented Sep 6, 2022 • edited by pr-preview bot Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

javifernandez commented Sep 12, 2022

javifernandez commented Sep 20, 2022

Choose a reason for hiding this comment

domenic left a comment • edited Loading

Choose a reason for hiding this comment

domenic left a comment

Choose a reason for hiding this comment

javifernandez commented Nov 22, 2022

domenic commented Nov 23, 2022

javifernandez commented Nov 23, 2022

javifernandez commented Dec 7, 2022

domenic commented Dec 7, 2022

javifernandez commented Jan 10, 2023

jgraham commented Jan 10, 2023

javifernandez commented Sep 6, 2022 •

edited by pr-preview bot

Loading

domenic left a comment •

edited

Loading