Fix datetime test.

Fix datetime test.

R=lrn@google.com

Committed: https://github.com/dart-lang/sdk/commit/581392cf17f77dcce6c9ee3112e19aa1f91993b3

[floitsch, 2015-12-09 06:19:29]

floitsch@google.com changed reviewers:
+ lrn@google.com

[Lasse Reichstein Nielsen, 2015-12-09 14:54:29]

lgtm

https://codereview.chromium.org/1511773003/diff/1/tests/corelib/date_time_tes...
File tests/corelib/date_time_test.dart (right):

https://codereview.chromium.org/1511773003/diff/1/tests/corelib/date_time_tes...
tests/corelib/date_time_test.dart:45: if (!supportsMicroseconds) return;
Still don't get why the test above works if microseconds are not supported, but
the one below does not.
In both cases you are creating a new DateTime from the microsecondsSinceEpoch of
another DateTime object, which should always give equal DateTime objects.

[floitsch, 2015-12-09 19:24:17]

https://codereview.chromium.org/1511773003/diff/1/tests/corelib/date_time_tes...
File tests/corelib/date_time_test.dart (right):

https://codereview.chromium.org/1511773003/diff/1/tests/corelib/date_time_tes...
tests/corelib/date_time_test.dart:45: if (!supportsMicroseconds) return;
On 2015/12/09 14:54:29, Lasse Reichstein Nielsen wrote:
> Still don't get why the test above works if microseconds are not supported,
but
> the one below does not.
> In both cases you are creating a new DateTime from the microsecondsSinceEpoch
of
> another DateTime object, which should always give equal DateTime objects.

Added a comment.

[floitsch, 2015-12-09 19:24:44]

Description was changed from

==========
Fix datetime test.
==========

to

==========
Fix datetime test.

R=lrn@google.com

Committed:
https://github.com/dart-lang/sdk/commit/581392cf17f77dcce6c9ee3112e19aa1f91993b3
==========

[floitsch, 2015-12-09 19:24:45]

Committed patchset #2 (id:20001) manually as
581392cf17f77dcce6c9ee3112e19aa1f91993b3 (presubmit successful).

[Lasse Reichstein Nielsen, 2015-12-10 08:35:05]

Still don't see how the test can fail, comment doesn't help.

You are testing the microsecondsSinceEpoch value of two DateTime instances where
the latter is created using DateTime.fromMicrosecondsSinceEpoch with the value
of the first.

No matter what, either DateTime truncates the value for the first instance, in
which case both values read using microsecondsSinceEpoch will be truncated, or
the first will not truncate - and then so won't the second.

The only way this can fail is if DateTime.now creates a value with microseconds
and microsecondsSinceEpoch returns that, but DateTime.fromMicrosecondsSinceEpoch
then truncates that value.

[floitsch, 2015-12-10 18:46:34]

On 2015/12/10 08:35:05, Lasse Reichstein Nielsen wrote:
> Still don't see how the test can fail, comment doesn't help.
> 
> You are testing the microsecondsSinceEpoch value of two DateTime instances
where
> the latter is created using DateTime.fromMicrosecondsSinceEpoch with the value
> of the first.
> 
> No matter what, either DateTime truncates the value for the first instance, in
> which case both values read using microsecondsSinceEpoch will be truncated, or
> the first will not truncate - and then so won't the second.
> 
> The only way this can fail is if DateTime.now creates a value with
microseconds
> and microsecondsSinceEpoch returns that, but
DateTime.fromMicrosecondsSinceEpoch
> then truncates that value.

That's what I was protecting against. Yes.
But you are right. We would need to go a bit into the future before this becomes
a problem.
Will update the test.