This adds coverage for the situation where we get a different answer from
Temporal.Duration.compare depending on whether relativeTo resolves to a
PlainDate or ZonedDateTime.
See discussion in
https://github.com/tc39/test262/pull/3505#discussion_r859994610
Some of these strings wouldn't have been valid even with a valid year in
them (e.g. strings ending in +01:00[UTC]) so fix up the strings that we
test. While touching these tests, I took the opportunity to regularize
them, and add some missing ones for ISO strings that convert to Calendar
and TimeZone.
Be consistent about creating Temporal objects for use in tests with direct
constructor calls, instead of relying on string coercions. This reduces
coupling in the tests.
Everywhere an ISO string is accepted in Temporal, a seconds value of :60
should always be coerced to :59, because of how leap seconds are handled
in ISO strings.
In property bags, a 'seconds: 60' property is not subject to that rule: it
should be handled according to the overflow option if there is one.
These tests existed already for some types; regularize them and add the
ones that didn't exist yet.
Each from() method except Calendar and TimeZone should test that when you
pass an instance of that type, the return value is a clone of that
instance, and a distinct object.
These tests existed already for some types; regularize them and add the
ones that didn't exist yet.
In order to test the referenceISODay of a PlainYearMonth we add an
argument to TemporalHelpers.assertPlainYearMonth.
In Date.parse() and new Date(), representations of the year 0 as -000000
must not be accepted. In the case of Date.parse(), they should yield NaN,
and in the case of new Date(), they should yield an invalid Date object,
whose valueOf() is NaN.
These tests cover, for every API entry point where a Temporal object is
expected, what happens when a value of a different type is passed in that
can't be converted.
Most entry points can convert a string to the expected Temporal type, and
will do ToString on any non-Object argument, and throw RangeError if the
result isn't a string that's convertible to that Temporal type. ToString
will throw TypeError on a Symbol.
Most entry points also take a property bag, and will throw TypeError if
the property bag doesn't have the required properties.
We also have to test for TimeZone and Calendar what happens if the wrong
type is provided as the value of a 'timeZone' or 'calendar' property in
another property bag, up to one level of nested properties.
Adds tests for conversion of a Number whose corresponding toString() value
is a valid ISO string. For some Temporal types this is possible, with a
number like 20220418.
Especially for Temporal.Calendar, we have to take into account the case
where the number is provided as the value for the 'calendar' property in a
property bag, and the case of up to one level of nested property bag as
well.
Regularizes and expands existing tests for this case.
The test for a TimeZone created from an ISO string with multiple offsets
was missing from the Temporal.ZonedDateTime constructor, whereas it was
present for several other APIs. Add it.
We already had similar tests to these for other methods, such as
Temporal.PlainTime.prototype.equals(). since() and until() should have
these tests too.
Update assertion messages in all of the existing tests as well, as per
Ms2ger's review comment.
An object may be given in any context where a Temporal.Instant is expected
(see ToTemporalInstant). There is no conversion from a property bag to an
Instant, unlike with other Temporal types. Instead the object is converted
to a string, and if its toString() method yields a valid ISO string, the
conversion succeeds. (An object with the default Object.prototype.toString
will not.)
We should make sure that we are providing the correct arguments to these
methods even if they are supposed to throw; they should throw for the
reason we expect, and not because we provided the wrong arguments.
Note that this test currently unintentionally passes, because a
TypeError is thrown for failing to convert the undefined returned from
the getter to a BigInt. But since this test was intended to test
detaching, it's no longer valid and should still be removed.
See https://github.com/tc39/test262/pull/3465#issuecomment-1098388916
The round() and toString() methods of Temporal.Instant, PlainDateTime, and
ZonedDateTime can round up or down. However, the instance must not be
treated as "negative" even when the time is before 1 BCE (years are
negative) or before the Unix epoch (epoch nanoseconds are negative). That
is, rounding down is always towards the Big Bang, and rounding up is
always away from it. Add tests that verify this.
This takes the tests of the rounding functionality of
Temporal.PlainTime.p.toString() and adds similar tests covering the
equivalent functionality to Duration, Instant, PlainDateTime, and
ZonedDateTime: all the types that have rounding and precision controls
for how they output their subsecond values.
It also takes the opportunity to improve the existing PlainTime tests:
- fractionalseconddigits-auto.js: More descriptive variable names. Added
assertion messages.
- fractionalseconddigits-number.js: Ditto.
- rounding-cross-midnight.js: Use constructor directly to remove coupling
with from().
- roundingmode-*.js: Add additional tests for specifying the precision
using fractionalSecondDigits.
- smallestunit-fractionalseconddigits.js: Add assertion messages.
This consolidates the few existing tests for options bags in Temporal
being of the wrong type, and adds them for every entry point in Temporal
that accepts an options bag.
These are mostly identical tests, but there is a variation for methods
like round() where either an options bag or string is accepted.
Add a consistent set of invalid strings for all of the
smallestunit-invalid-string.js and largestunit-invalid-string.js tests:
- "era" and "eraYear" in singular and plural
- all of the units that are disallowed for that particular method call, in
singular and plural
- an allowed unit with \0 at the end
- an allowed unit with an "i" replaced by a dotless i
- an allowed unit but with all-caps
- an unrelated string.
Of the toString() methods that have options for printing a time with
seconds and fractional seconds, PlainTime seems to have the most
comprehensive set of tests. Bring all the others (Duration, Instant,
PlainDateTime, and ZonedDateTime) in sync with PlainTime, and edit the
PlainTime ones where necessary to include improvements from the others.
Tests:
- fractionalseconddigits-invalid-string.js: copy and expand on
PlainTime's more comprehensive set of invalid strings. Add assertion
message. Fix front matter.
- fractionalseconddigits-non-integer.js: Fix front matter.
- fractionalseconddigits-out-of-range.js: make sure infinity is tested.
Add assertion messages. Fix front matter.
- fractionalseconddigits-undefined.js: copy PlainTime's more
comprehensive test with whole minutes, whole seconds, and subseconds.
Copy PlainTime's test of an empty function object. Add more
descriptive variable names and assertion messages. Fix front matter.
- fractionalseconddigits-wrong-type.js: inline and delete TemporalHelper
used here; it was only good for this test anyway. Improve assertion
messages.
- smallestunit-valid-units.js: copy PlainTime's test with a second value
with zero seconds even. Refactor repetitive tests into a loop. Copy
the invalid unit "era" from the Instant test. Add assertion messages.
A prior version of ECMA262 described invalid mathematical operations
with infinite values [1]. Update the test metadata to reflect the
corrected specification text, and add two assertions for the obsolete
conditions.
[1] ".bind on a function with infinite length has imprecise spec and
engine divergences"
https://github.com/tc39/ecma262/issues/2170
Philip Chimento
Ms2ger
Jesse Alama
Jordan Harband
Frank Tang
rwaldron
Jesse Alama
Shu-yu Guo
Jesse Alama
Mike Pennisi