Hi Emmanuel. Thanks for the comments! I've also applied them to
jackson-datatype-guava and jackson-module-afterburner as appropriate.
On 5/11/14 7:26 PM, "Emmanuel Bourg" <ebourg@apache.org> wrote:
>- do not hesitate to create your repositories for Java packages under
>the pkg-java group on alioth. You'll find a setup-repository script on
>alioth in the /srv/git.debian.org/git/pkg-java directory that prepares
>the repository and wires the receive hooks properly (with mail and IRC
>notifications). Also if you set the maintainer in debian/control to the
>'Debian Java Maintainers' this will allow us to co-maintain the package.
I've moved everything over to pkg-java now so all development should
happen there now and in the future.
>- debian/control: ${shlibs:Depends} is not necessary for lib*-java
>packages
Fixed.
>- debian/control: in general ${maven:OptionalDepends} is best affected
>to the Suggests field than Recommends. jackson-datatype-joda has no
>optional dependency so there is no impact here.
Fixed.
>- adding a short description to the patches is usually a good idea (see
>http://dep.debian.net/deps/dep3/). It consists in a 'Description' field
>explaining the purpose of the patch and why it was needed, an 'Author'
>field with your name and a 'Forwarded' field to track if the patch has
>been contributed back upstream.
I've added DEP3 headers to all patches.
>- debian/maven.rules: The artifacts are installed with the 2.x generic
>version. Usually it's preferable to use the 'debian' version unless
>there is a conflict with another package. Otherwise when the version is
>bumped to 3.0, you either stay with 2.x which is odd, or update to 3.x
>but have to modify the rules in the reverse dependencies.
Not fixed, as mentioned in your subsequent email.
>- debian/rules doesn't export JAVA_HOME, this is important to ensure the
>default-jdk is used and not the current Java alternative as this may
>produce code incompatible with the default Java.
Fixed.
>- I got a test failure:
>
>Failed tests:
>testDateMidnightSer(com.fasterxml.jackson.datatype.joda.JodaSerializationT
>est):
>expected:<"2001-05-2[5]"> but was:<"2001-05-2[4]">
>
>testDateMidnightSerWithTypeInfo(com.fasterxml.jackson.datatype.joda.JodaSe
>rializationTest):
>expected:<...Midnight","2001-05-2[5]"]> but
>was:<...Midnight","2001-05-2[4]"]>
>
>This is an issue related to the timezone, the test fails on a system
>with Europe/Paris, but succeeds with GMT.
I can reproduce this but couldn't figure out whether it's just an external
dependency in the test suite or a genuine bug in the code. I've added an
"export TZ=GMT" to debian/rules to make the tests pass and filed an
upstream bug to see if they have any comment about it.
Regards,
Tim.
>
>Emmanuel Bourg
>
>
Attachment:
smime.p7s
Description: S/MIME cryptographic signature