Alexei Barantsev Reporter
	Description • 6 years ago

User Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.108 Safari/537.36

Steps to reproduce:

Detected in Firefox Nightly 59.0a1 (2018-01-14) (64-bit)

Regression issue, may be caused by https://bugzilla.mozilla.org/show_bug.cgi?id=1354211

driver.get("https://output.jsbin.com/serilo");
WebElement numInput = driver.findElement(By.id("num"));
numInput.sendKeys("test");
numInput.clear();
numInput.sendKeys("33");
assertEquals(numInput.getAttribute("value"), "33");


Actual results:

getAttribute returns empty value, but visually the field contains text "test33"


Expected results:

getAttribute should return value "33"

	Henrik Skupin [:whimboo][⌚️UTC+1]
	Comment 1 • 6 years ago

Andreas, can you please have a look?

	Andreas Tolfsen ❲:ato❳ Assignee
	Comment 2 • 6 years ago

This appears to be a bug in the WebDriver standard.  We abort
modifying the element if it is empty, which for an <input> is if its
value property is equal to "" (empty string).

The complication is that whilst it is allowed to enter invalid
characters to an <input type=number> in Gecko, it is not in Chrome.
This means this particular WebDriver bug will only occur in Gecko
because the value property isn’t updated when the input is
invalid.

As far as I know there is no way with current web platform APIs to
access the current (invalid) input of <input type=number>.  But we
can additionaly make a check of its ValidityState [1], since the
value === "" check is not sufficient.

I filed [2] against the WebDriver specification.

Regarding the above test case, you should be getting the input’s
_property_ rather than the value, but I corrected this in my own
test case.

  [1] https://html.spec.whatwg.org/#validitystate

	David Burns :automatedtester
	Comment 3 • 6 years ago

[Mass update] Setting P1 as currently being worked on.

	Alexei Barantsev Reporter
	Comment 4 • 6 years ago

I want to make a note that it works in the current Release (57) and used to work in 59 before the mentioned change https://bugzilla.mozilla.org/show_bug.cgi?id=1354211. So I suppose it is technically possible to implement "old" behaviour. But, yes, it should be coherrent with the specification.

	Andreas Tolfsen ❲:ato❳ Assignee
	Comment 9 • 6 years ago

try because try-from-mozreview is broken: https://treeherder.mozilla.org/#/jobs?repo=try&revision=f4e3e746960f0d2a3e65f3f076afaa412b7ad810

	David Burns :automatedtester
	Comment 17 • 6 years ago
mozreview-review

Comment on attachment 8942692 [details]
Bug 1430575 - Adjust editable definition to match WebDriver.

https://reviewboard.mozilla.org/r/212954/#review218654

	David Burns :automatedtester
	Comment 18 • 6 years ago
mozreview-review

Comment on attachment 8942693 [details]
Bug 1430575 - Document interaction.clearElement.

https://reviewboard.mozilla.org/r/212956/#review218656

::: testing/marionette/interaction.js:295
(Diff revision 2)
>  };
>  
> +/**
> + * Clears the form control or the editable element, if required.
> + *
> + * Before attempting to clear the element, it will be attempted

s/attempted scrolled into view/attempt to scroll into view

::: testing/marionette/interaction.js:300
(Diff revision 2)
> + * Before attempting to clear the element, it will be attempted
> + * scrolled into view if it is not already in the viewport.
> + *
> + * If the element is a submittable form control and it is empty
> + * (it has no value or it has no files associated with it, in the
> + * case it is a <code>&lt;input type=file>&gt; element) or it is an

closing `</code>` is missing which could explain the `>` next to `type=file`

	David Burns :automatedtester
	Comment 19 • 6 years ago
mozreview-review

Comment on attachment 8942694 [details]
Bug 1430575 - Add element.satisfiesContraints.

https://reviewboard.mozilla.org/r/212958/#review218664

::: testing/marionette/element.js:964
(Diff revision 2)
> + *     True if it satisfies all its constraints.
> + *
> + * @throws {InvalidElementStateError}
> + *     If <var>el</var> is not a mutable form control element.
> + */
> +element.satisfiesConstraints = function(el) {

I don't like this this method name as it looks like it should be used as `element.satisiesConstraints(el, constraint);`

As the method is so small why can't this just be in method where you call `element.satisfiesConstraints(el);` in your other patch

	Andreas Tolfsen ❲:ato❳ Assignee
	Comment 20 • 6 years ago
mozreview-review-reply

Comment on attachment 8942694 [details]
Bug 1430575 - Add element.satisfiesContraints.

https://reviewboard.mozilla.org/r/212958/#review218664

> I don't like this this method name as it looks like it should be used as `element.satisiesConstraints(el, constraint);`
> 
> As the method is so small why can't this just be in method where you call `element.satisfiesConstraints(el);` in your other patch

I agree.  Merging it with clearResettableElement in
testing/marionette/interaction.js makes the code flow much nicer.

	David Burns :automatedtester
	Comment 24 • 6 years ago
mozreview-review

Comment on attachment 8942695 [details]
Bug 1430575 - Add validity state check for WebDriver:ElementClear.

https://reviewboard.mozilla.org/r/212960/#review219316

	Pulsebot
	Comment 25 • 6 years ago

Pushed by atolfsen@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/2f1762f3233a
Adjust editable definition to match WebDriver. r=automatedtester
https://hg.mozilla.org/integration/autoland/rev/7220cf5de9a8
Document interaction.clearElement. r=automatedtester
https://hg.mozilla.org/integration/autoland/rev/949cf88764e6
Add validity state check for WebDriver:ElementClear. r=automatedtester

	Tiberius Oros[:tiberius_oros]
	Comment 26 • 6 years ago
bugherder

https://hg.mozilla.org/mozilla-central/rev/2f1762f3233a
https://hg.mozilla.org/mozilla-central/rev/7220cf5de9a8
https://hg.mozilla.org/mozilla-central/rev/949cf88764e6