URLSearchParams constructor interprets an array as a record

[saschanaz]

const params = new URLSearchParams(["ab"]);
document.write(params);

Expected: An error, because the constructor requires sequence<sequence<USVString>>.
Actual: Instead it interprets as a record, so the result is 0=ab&length=1. The spec says sequences and records are distinguishable, so this shouldn't happen.

See also #19776 which is also a distinguishablity problem.

[saschanaz]

@highfive assign me

BTW, is it expected that the bot just adds a label without actually adding me as an assignee?

[highfive]

Hey @saschanaz! Thanks for your interest in working on this issue. It's now assigned to you!

[CYBAI]

BTW, is it expected that the bot just adds a label without actually adding me as an assignee?

Yes, highfive bot will only add the C-assigned label because GitHub didn't support assigning someone out of the organization when the bot was created.

[CYBAI]

Btw, maybe it's good to add the testcase you gave into this one 👀?

servo/tests/wpt/web-platform-tests/url/urlsearchparams-constructor.any.js

Lines 150 to 158 in 1f64024

	test(function() {
	var params = new URLSearchParams([]);
	assert_true(params != null, 'constructor returned non-null value.');
	params = new URLSearchParams([['a', 'b'], ['c', 'd']]);
	assert_equals(params.get("a"), "b");
	assert_equals(params.get("c"), "d");
	assert_throws(new TypeError(), function() { new URLSearchParams([[1]]); });
	assert_throws(new TypeError(), function() { new URLSearchParams([[1,2,3]]); });
	}, "Constructor with sequence of sequences of strings");

[saschanaz]

This is a general binding problem so it's possible that there already is a failing test elsewhere. I might check after implementing a fix.