Backporting profiler related changes into Apache NetBeans.

[JaroslavTulach]

VisualVM project decided to pursue its own development of profiler.lib and profiler.oql since affe1eb1308554. Understandable given the complicated situation NetBeans project was in at that time. However since then the Apache NetBeans project has proven its maturity, ability to deliver and produce regular quarterly releases, as well as upload all the NetBeans libraries to Maven central. Maven central is the default way of distributing and consuming the libraries and I am reusing profiler.lib in Insight project. While doing so, I noticed slight differences in behavior and in order to mimize them I am proposing applying them to Apache NetBeans profiler.lib and profiler.oql libraries.

First of all I took the changes from

$ git diff affe1eb1308554...e88fe0edbb

and filtered them out to contain only the changes in above two libraries:

filterdiff -i '*/visualvm/libs.profiler/lib.profiler/**' | \
  sed s#visualvm/libs.profiler/lib.profiler#profiler/lib.profiler#g | \
  sed s#org/graalvm/visualvm/lib/jfluid#org/netbeans/lib/profiler#g`

then I applied the changes as a07981a and resolved remaining clashes manually. Now I am creating a pull request to let the integration tests run and notify me of remaining issues that need to be polished.

[neilcsmith-net]

-1 for now - this surely raises some licensing issues?! cc. @matthiasblaesing @geertjanw

[matthiasblaesing]

The big question is: Who is the author of the code and in who owns copyright/can relicense the code. Options I can speculate about:

• @JaroslavTulach is the author himself, then everything is fine, he is a committer, has therefore an ICLA on file
• all authors are committers/have an ICLA on file or only provided minimal changes, then I still think we can merge
• the work was done under the Oracle OCLA umbrella, then contribution could be covered by an CCLA (not sure whether OCLA and CCLA are transitive) or a Software License Grant
• something other

In any case I like the intention to get this all merged again so in general a +1 for the intention and a -1 for the unclear license question.

[JaroslavTulach]

No, the code wasn't written by me. I tried to admit that by making myself a committer, but not author of a07981a. Deeper investigation of all the commits shows:

visualvm$ git log affe1eb1308554...e88fe0edbb | grep ^Author: | sort -u
Author: Brandon Fish <brandon.j.fish@oracle.com>
Author: Jiri Sedlacek <jiri.sedlacek@oracle.com>
Author: Tomas Hurka <tomas.hurka@gmail.com>
visualvm$ git log affe1eb1308554...e88fe0edbb | grep ^Author.*Hurka | wc -l
552
visualvm$ git log affe1eb1308554...e88fe0edbb | grep ^Author.*Sedlacek | wc -l
446
visualvm$ git log affe1eb1308554...e88fe0edbb | grep ^Author.*Fish | wc -l
1

E.g except oracle/visualvm@9efce02 all the code is written by either @thurka or @jisedlac. Now, let's try to select just the commits related to visualvm/libs.profiler/lib.profiler. I have written following filter.sh script:

git log affe1eb1308554...e88fe0edbb | grep commit | cut -d " " -f2 | while read X; do 
  D=`git show $X | filterdiff -i '*/visualvm/libs.profiler/lib.profiler/**'`
  if [ "x$D" == "x" ]; then 
     # echo None $D
     IGNORE=$D
   else 
     git show $X | grep ^Author: 
   fi 
done

Running filter.sh | wc -l yields 87. 78 commits by @thurka and 9 by @jisedlac. The oracle/visualvm@9efce02 commit isn't included in a07981a at all.

[matthiasblaesing]

Thank you for the analysis. So could the two main authors get an ICLA on file? I looked through the ICLA signee list and did not find either.

[neilcsmith-net]

@matthiasblaesing well, we don't strictly need an ICLA for contributions even if useful - only committers require one. But either way, we surely need the clear intention of the authors (and/or employer if necessary) to re-license and contribute these particular changes here?

[matthiasblaesing]

@neilcsmith-net ah yes - you are right. To have some permanent record, a tracking issue in JIRA would then be helpful.

[JaroslavTulach]

Yes, having consents from both authors is a pre-requisitte to integrate. I haven't obtained the permission so far. Closing. Whether there is a way to avoid doing the same work twice remains to be seen.