Review request for JDK-8016760: failure of regression test langtools/tools/javac/

Xueming Shen xueming.shen at
Tue Jun 25 18:47:50 UTC 2013

It is intended to be an opinion. I would assume you would like a reviewer from
the langtool team to push the change, given the changset is for the lang repo.

You may want to update the "check()" method as well to low the granularity
to 2-second before comparing the equality. The timestamp from the ms-dos
formatted date/time has 2-second granularity, but the new one has second
granularity, so the test case may still fail in some odd timing.


On 06/25/2013 11:33 AM, Eric McCorkle wrote:
> Is this a capital-R review?
> On 06/25/13 13:50, Xueming Shen wrote:
>> This is fine to be a workaround for the test case for now. It probably
>> will need to be
>> undo-ed after the propose change for #8015666 get integrated.
>> The proposal for #8015666 is to keep the "existing" behavior of
>> ZipEntry.getTime()
>> to return a LastModifiedTime converted from the zip entry's
>> ms-dos-formatted date/time
>> field by using the "default" timezone. A new pair
>> ZipEntry.get/setLastModifiedTime()
>> will be added to access the "real" UTC time stored in the zip entry, if
>> presents.
>> The API doc will be updated accordingly as well to explicitly explain
>> the source of the
>> date/time and the its timezone sensitive conversion.
>> -Sherman
>> On 06/25/2013 07:03 AM, Eric McCorkle wrote:
>>> Hello,
>>> Please review this simple patch which updates regression test
>>> langtools/tools/javac/ to offset the time returned by
>>> JavaFileObject.getLastModified() with the local time to UTC delta.
>>> Please note that this patch is intended to address the test failures,
>>> and that I will be immediately opening a new bug to investigate and
>>> address deeper issues, and also to properly document the API.
>>> The webrev is here:
>>> The bug report is here:
>>> Thanks,
>>> Eric

More information about the core-libs-dev mailing list