← Back to team overview

openjdk team mailing list archive

Bug#944738: Bug#944738: jlink: Hash of module differs to expected hash recorded in java.base

 

On Wed, Sep 23, 2020 at 08:55:11PM +0100, Julian Gilbey wrote:
> On Wed, Sep 23, 2020 at 11:09:28AM -0700, tony mancill wrote:
> > > dh_strip_nondeterminism gets its timestamp from the debian/changelog
> > > file, while strip-nondeterminism strips the timestamps.
> > > 
> > > So here's a replacement patch which fixes everything except
> > > java.naming.jmod.  I have to run now, but will try to figure out
> > > tomorrow why that one still breaks.
> > 
> > I rebuilt with this patch (and also my NONPARALLEL patch) and am able to
> > call jlink against all of the modules shipped with the resulting JDK
> > binary package.  So this is looking good.
> > 
> > I'm rebuilding now (without nonparallel) what should be the NMU candidate.
> 
> Super!  I wonder, though, why it didn't work for me first time round?
> I rebuilt it and it worked the second time, though.  So maybe I
> changed something.

The rebuild of openjdk-14, including tests and the ad-hoc jlink tests
was successful, so I think this is ready.  I have uploaded it to
DELAYED/15, closing #968991 [1], since there was a bug there for just
openjdk-14.

> Obviously, similar builds will be needed for openjdk-{11,13,14,15}.

Yes, I will start working on the others.

Thank you for all of help.
tony

[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=968991)


References