From: David P G. <gr...@us...> - 2011-08-22 20:31:29
|
"Thomas Preud'homme" <ro...@ce...> wrote on 08/20/2011 04:11:08 PM > > > 1/ About copyrights > > a) The directory docs/images contains several images, some of them > being under > CPL license (docs/images/fat-jalapeno-pepper-cropped-*.gif) > according to their > meta-information. Since there is no exception for this directory in > LICENSE.html, are they to be considered all under EPL 1.0? It is my > understanding that CPL is a bit more restrictive than EPL, hence if the file > matching the pattern docs/images/fat-jalapeno-pepper-cropped-*.gif are really > under CPL, I guess the whole archive is to be redistributed under CPL. The copyright on those images is owned by IBM. When we re-licensed all of Jikes RVM from CPL to EPL, I guess we missed some of the meta-data. If you can tell me how to fix the meta-information to reflect that the images are licensed under the EPL, I will do it. > > b) Several files in testing/tests seems non-free to me, namely: > > - testing/tests/SPECjbb*/*.props > - testing/tests/perf-jbb*/*.props > - testing/tests/pseudojbb/* > - testing/tests/javalex/qb1.lex.ref.classpath > > The first three seems to come from SPECjbb and not written by Jikes RVM > contributors. But from what I understand, SPECjbb is non free so I'mnot sure > it's fine (including for you) to distribute these files along Jikes RVM. > > The fourth and last item is copyrighted: > > 1999-2000 Toshihiro Horie > 1999 Regents of the University of California > > On the webpage of Toshihiro Horie [0], we can read: > > "I will make more of the sources available as this project > progresses. All of > the source and specification files are copyrighted by Toshihiro Horie and UC > Regents. They may be modified or used only via the written consent of the > author. However, the final project is planned to be released under > the terms of > the GNU Public License." > > [0] http://www.ocf.berkeley.edu/~horie/qbjc.html: For Debian, perhaps the thing to do is for you to cut these directories out of the package. > > c) I wouldn't bother you about this if it was the only remark but since I'm > already writing this email… In the directory userguide/, only the root > index.html contains the EPL header. Do you confirm all of the files > are covered > by EPL? (I just need a ack on this) > Yes. they are under the EPL. They are generated by exporting html from our wiki. It might be possible to figure out how to include a EPL header in them as part of the export process, but the lack of a license header in the file is not intended to be significant. > > 2/ About dependencies > > a) Jikes RVM build system expect exact version match to build. Is this exact > match in the versions really necessary or would a >= version be ok? If yes, > I'll push a patch on my machine and will come back to you with a link to it. > Not sure. May depend on the package. > > b) Jikes RVM currently build depend on Harmony or GNU classpath. Both of them > are not packaged in Debian and I'd like (if possible) to avoid > packaging them. > gcj is packaged though and it seems gcj and GNU classpath sources > are one now. > One solution would be to use one of the package gcj-{4.4,4.6}-source which > seem to contain GNU classpath. But I'd prefer Jikes RVM build to only build > Jikes RVM. > > It is my understanding though that GNU classpath is only used for building > glibj.zip (renamed as classpath.jar) and from that building rt.jar. I noticed > that the packages gcj-{4.4,4.6}-jre-lib contain a file rt.jar quite close to > the content of glibj.zip. Most of the classes are the same, except for 2 > additional *packages* in rt.jar, one of them being awt. > > c) My question is: would it work to use this rt.jar to replace both > classpath.jar and rt.jar built from GNU classpath? If yes, would > there be some > consequences? > Jikes RVM patches the libraries and also configures their native code as part of its build process. I would not expect Jikes RVM to work with a vanilla binary build of either GNU Classpath or Apache Harmony. Sorry. --dave |