Am 29.03.2017 um 15:30 schrieb Ole Streicher: > Hi Markus, > > Am 29.03.2017 um 15:15 schrieb Markus Koschany: >>> There is no jar without a version number. This makes it impossible to >>> add the jar to the CLASSPATH of other packages, >> >> This is not true. You can add such jar files to your CLASSPATH. >>> since the file name will change silently with the next version. >> >> This is true. > > This is my point: we cannot rely on it. > >>> Therefore, please include also a versionless jar file >>> BrowserLauncher2.jar. This is also requested by Java policy. >> >> Java Policy is more like Parley from Pirates of the Caribbean and >> incomplete by the way. I suggest to downgrade the bug to "important" >> because it does not affect jmodeltest, the only reverse-dependency at >> the moment and a missing symlink is not release critical in this case >> and can be trivially worked around. > > It affects the package: in the moment when a new version is uploaded (my > it be via backports), it will silently break jmodeltest. It will also > silently break local usage, even if the new version would be otherwise > compatible. Both is not acceptable for a release, hence the RC status. > > Best regards > > Ole
In the case of backports you would probably backport jmodeltest as well, so both could be easily kept in sync. AFAICT both packages seem to work well together and librowserlauncher-java was only packaged to build jmodeltest from source. Looking at the issue at hand I still think that you don't need to choose RC severity but I honestly don't care enough about it and just wanted to give some feedback. #859005 is probably more important. I would talk to Andreas about it directly (in case he missed it somehow). Regards, Markus
signature.asc
Description: OpenPGP digital signature
__ This is the maintainer address of Debian's Java team <http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-java-maintainers>. Please use debian-j...@lists.debian.org for discussions and questions.
