For bug 188007 we'd need a newer version of istack-commons, one which contains com.sun.istack.XMLStreamReaderToContentHandler which appeared in the istack CVS on 2007-02-22. I have no idea where the istack tarball on the Gentoo mirrors came from, probably straight out of CVS. Some comment about this would have been valuable. In any case, if annyone could bump that up, I'd appreciate that.
(In reply to comment #0) > For bug 188007 we'd need a newer version of istack-commons, one which contains > com.sun.istack.XMLStreamReaderToContentHandler which appeared in the istack CVS > on 2007-02-22. I have no idea where the istack tarball on the Gentoo mirrors > came from, probably straight out of CVS. Some comment about this would have > been valuable. In any case, if annyone could bump that up, I'd appreciate that. > nelchael did the ebuilds so he should hopefully be able to tell about the tarballs.
I found a way to get a directory that looks much like current tarballs, but it requires authenticated access to the CVS server. cvs -d :pserver:username@cvs.dev.java.net:/cvs login cvs -d :pserver:username@cvs.dev.java.net:/cvs \ checkout istack-commons/istack-commons istac-commons cd istack-commons/istack-commons find -name CVS -print0 | xargs -0 rm -r rm -rf lib/* test/lib/* runtime/lib/* tools/lib cd .. mv istack-commons{,-$(date +%Y%m%d)} If you are sure about the CDDL licensing, then redistributing this source should be no problem no matter what the CVS access looks like. However I could find no licensing information in that tree, so I'm not sure. On the other hand, the CVS viewer online is unrestricted, so this CVS access is probably only a config issue.
With a tarball created as described in comment 2 and ebuilds renamed accordingly, both istack ebuilds compiled all right, and the generated jar files contain all classes that were present before, and a few more as well.
Petteri Räty aka Betelgeuse kindly pointed out that guest/guest could be (and in fact is) a valid login for that CVS. In the meantime I found out that the last changes were 2007-02-22, after that there was only one modification with the toplevel build.xml which we don't use anyway. Therefore I suggest the following steps to generate the tarball: cvs -d :pserver:guest@cvs.dev.java.net:/cvs login cvs -d :pserver:guest@cvs.dev.java.net:/cvs \ checkout -D 2007-02-23 istack-commons/istack-commons cd istack-commons/istack-commons find -name CVS -print0 | xargs -0 rm -r rm -rf lib/* test/lib/* runtime/lib/* tools/lib cd .. mv istack-commons{,-20070223} tar cvjf istack-commons-20070223{.tar.bz2,}
There is also a jar file calles istack-commons-test, matching the directory test in the source tree. Probably there should be an ebuild for this as well. What's the rationale behind splitting the whole istack-commons into different ebuilds?
Created attachment 127215 [details, diff] Patch to update runtime and tools and add test I added ANT_TASKS="none" to all my ebuilds and thereby noticed a dependency in istack-commons-runtime on dev-java/jsr173. The istack-commons-test ebuild is based upon the istack-commons-tools ebuild, modifying the name where appropriate and adjusting for different dependencies. I'm not sure why istack-commons-tools symlinks java-config --tools, but I copied that as well.
(In reply to comment #5) > What's the rationale behind splitting the whole istack-commons into different > ebuilds? Not all packages need tools, simple? java-config --tools symlinks tools.jar from JDK. As to releases from dev.java.net: they suck, suck, suck. If they provide a tar/zip file - it may disappear any day, is being changed (and not renamed), etc. There's a lot of troubles with djn, so most things are CVS snapshots - it's the only safe method to be sure of the ability to recreate the archive.
No activity here in over a year. In the meantime, it looks like istack-commons has had a version 1.0 released this january. At least there is a tag called istack-commons-1_0 in their CVS, dated 2008-01-08. STill no source tarball to be found, so again I'd suggest pulling sources from their CVS. cvs -d :pserver:guest@cvs.dev.java.net:/cvs login cvs -d :pserver:guest@cvs.dev.java.net:/cvs \ checkout -r istack-commons-1_0 istack-commons/istack-commons It also looks like the new version could be built with maven - at least there are pom.xml files lying around, and some CVS log messages seem to come from maven. Do you want this release in portage?
(In reply to comment #8) > No activity here in over a year. In the meantime, it looks like istack-commons > has had a version 1.0 released this january. At least there is a tag called > istack-commons-1_0 in their CVS, dated 2008-01-08. STill no source tarball to > be found, so again I'd suggest pulling sources from their CVS. > Consider getting access to our overlays as having the ebuilds there makes it easier for us to move things. Just came by #gentoo-java@freenode and I can arrange it.
+*istack-commons-runtime-2.22 (12 Jun 2015) + + 12 Jun 2015; Patrice Clement <monsieurp@gentoo.org> + +istack-commons-runtime-2.22.ebuild: + Version bump. Fix bug 188015. +
+*istack-commons-tools-2.22 (12 Jun 2015) + + 12 Jun 2015; Patrice Clement <monsieurp@gentoo.org> + +istack-commons-tools-2.22.ebuild: + Version bump. Fix bug 188015. +
+*istack-commons-buildtools-2.22 (12 Jun 2015) + + 12 Jun 2015; Patrice Clement <monsieurp@gentoo.org> + +istack-commons-buildtools-2.22.ebuild: + Version bump. Fix bug 188015. +
+*istack-commons-soimp-2.22 (12 Jun 2015) + + 12 Jun 2015; Patrice Clement <monsieurp@gentoo.org> + +istack-commons-soimp-2.22.ebuild, +metadata.xml: + Initial commit. Fix bug 188015. v2 introduced a new module which has now been added to the tree. Closing this bug.