openjdk team mailing list archive
-
openjdk team
-
Mailing list archive
-
Message #14065
Bug#944738: Bug#944738: jlink: Hash of module differs to expected hash recorded in java.base
* Julian Gilbey:
> On Fri, Sep 25, 2020 at 06:22:33PM +0200, Florian Weimer wrote:
>> * tony mancill:
>>
>> > I had already prepared the NMU for openjdk-15 - building first without
>> > the patch and then with it so I could test with both - so I will go
>> > ahead and upload that package today.
>>
>> There's also an upstream change to honor SOURCE_DATE_EPOCH:
>>
>> commit 1a16a4b62829e54a4af5446ef15b981212a70718
>> Author: Magnus Ihse Bursie <ihse@xxxxxxxxxxx>
>> Date: Thu May 7 17:47:11 2020 +0200
>>
>> 8244592: Start supporting SOURCE_DATE_EPOCH
>>
>> Reviewed-by: erikj
>>
>> It requires configuring with --with-source-date, I think.
>
> Oh, interesting. That could be very helpful for this. But we would
> have to be careful, as dh_strip_nondeterminism replaces timestamps
> with the date in the most recent debian/changelog entry, so we might
> have to exclude the relevant files from the dh_... invocation.
Maybe setting SOURCE_DATE_EPOCH to the same date will work?
References