[ANNOUNCE] Java Git (aka jgit) has switched to 3-clause BSD

As of 53a2cc3 the jgit library (a 100% pure Java implementation of git) is now licensed under a 3-clause (new-style) BSD license. The change was done with a Perl script to rewrite all source code headers within the org.spearce.jgit package. Copyright information was updated based upon the output of git-blame.

The change is here:

	git://repo.or.cz/egit/spearce.git bsd

and is based upon the the thread that I started on this list:

	http://article.gmane.org/gmane.comp.version-control.git/81585

If there are no objections within the next couple of days we'll merge it to the main tree.

Hi,

On Mon, May 26, 2008 at 6:46 AM, Shawn O. Pearce <spearce@spearce.org> wrote:

I previously read about this transition from GPL to BSD.

IANAL, but isn't the current version a derivative of the older version, meaning that you need to take all authors into account when doing this switch, and not just the authors of the _current_ code?

(Or did you get a consent from all previous authors as well as current authors?)

Vegard

Hi,

On Mon, 26 May 2008, Vegard Nossum wrote:

He did. That is what is written in the thread that Shawn explicitely stated in the part that you did _not_ quote.

I also have to wonder why you chose a total technical non-issue, that does not really concern you (because you are noone of said authors), for your first post (at least that I am aware of) to this list.

Ciao, Dscho

Hi,

On Mon, May 26, 2008 at 12:08 PM, Johannes Schindelin <Johannes.Schindelin@gmx.de> wrote:

Hm, yes, that is the thread I read before. What Shawn writes there is this:

"As of the bleeding edge (40c5c6cb11b8cc6caf3ea6a681caf0a6b8d66f36 [*4*]) the ownership of all currently surviving lines of code is broken down as follows:"

This seems to suggest that only authors of the "bleeding edge" are considered, while my point was that the "bleeding edge" may be a derivative of earlier versions which had other authors as well.

I can't find anywhere explicitly mentioned in the thread that _all_ authors have been asked about the change of license.

It does concern me, because I enjoy git and free software in general. I am not trolling or trying to stop the change of license in any way, but simply give a heads up to what I believe might potentially be a legal issue. Maybe there are in fact no other authors that have participated, but in that case I think it is a fact worthy of explicit mention.

This was also not my first post to this list; you have even replied to one of my posts yourself once upon a time :-)

Vegard

Hi,

On Mon, 26 May 2008, Vegard Nossum wrote:

AFAICT this was done with "git blame", so it might be "derived" by a particularly SCO-esque interpretation of the word, but it is really the work of those people Shawn mentioned.

Besides, I do not think that there were any other authors anyway.

In any case, this discussion is already boring me mightily, so I will tune out, Dscho

On Mon, May 26, 2008 at 01:22:25PM +0200, Vegard Nossum wrote:

I share your concerns about licensing and keeping things accurate when it comes to things like changing licenses, so I took a quick look at the bsd branch of the jgit git tree, to see if there were any glaring omissions.

I'm not sure that git-blame is the best way to deal with this kind of task, but in this case, it worked out well. From my analysis of git-log output, the lines of code from contributors who haven't explicitly consented are maybe 10 lines. Looks like a job well done to me.

- Chris

Chris Frey <cdfrey@foursquare.net> wrote:

Thanks for your comments. I think we are OK with the change, as Chris points out. Here's more details about the complete list of contributors to jgit, since the beginning of time.

$ git log --pretty=format:%ae \
    bsd jgit org.spearce.jgit org.spearce.jgit.test \
    | sort | uniq -c | sort -nr
 364 spearce@spearce.org
 175 robin.rosenberg@dewire.com
  21 dwatson@mimvista.com
   8 me@lathund.dewire.com
   2 rogersoares@intelinet.com.br
   1 thadh@thad.corp.google.com

Now I asked for consent from the top 3 (me, robin, dwatson). The 4th one down (me@lathund.dewire.com) is also Robin. That leaves only Roger C. Soares and Thad Hughes as ever having made a contribution to jgit, surviving line or not, and who were not consulted specifically about the license change.

The total size of their deltas is:

$ for u in rogersoares@intelinet.com.br thadh@thad.corp.google.com; \
  do \
    echo $u $(git log -p --author=$u bsd \
    jgit org.spearce.jgit org.spearce.jgit.test \
    | grep '^+' | wc -l); \
  done
rogersoares@intelinet.com.br 9
thadh@thad.corp.google.com 5
$ git log -p bsd | grep '^+' | wc -l
  122576

So in all time these users have added a total of 14 lines between them. Out of 122,576 total added lines. It amounts to only 0.011% of the total contribution. In the US a change this small may not even be copyrightable.

Obviously I am only counting + lines in patches. The - lines may also have some value in certain contexts. But then we also get into the discussion of what is copyrighted. The patch or the final result?

In the 3 changes between Roger and Thad the patches were all very small. They are minor bug fixes that probably would come out the same way no matter who wrote them. E.g. in one Roger deleted an extra ";" at the end of a line (fixed ";;" to ";"). I'm not sure anyone else on Earth could have written that change differently.

I hate to trivialize someone's contribution, but the reality is that in jgit there are only 3 interesting authors at this time. All were consulted about this license change, and all gave their consent via a public posting to this mailing list.

On Tue, May 27, 2008 at 11:44 AM, Shawn O. Pearce <spearce@spearce.org> wrote:

It'd be nice to hear from them, but I looks fair to me to say that those contributors - while very valuable - cannot hold the project back.

Don't personally use jgit or Eclipse but the porting projects (jgit, msysgit) are fantastic. Thanks!

m

Martin Langhoff escreveu:

Well, FWIW I'm ok with the license change.

[]s, Roger.

"Roger C. Soares" <rogersoares@intelinet.com.br> wrote:

Heh, thanks for chiming in Roger. We're likely to start talking about moving egit to 100% EPL soon too. I think your contribution there is quite non-trivial, so we'll certainly need your blessing for that. :-)

Shawn O. Pearce escreveu:

Hi Shawn, Robin told me about it and asked for permission some time ago, so my permission is given already. Anyway, I can confirm this later if necessary. :)

[]s, Roger.

"Shawn O. Pearce" <spearce@spearce.org> writes:

Errr... why bot use git-shortlog?

Hi Shawn

will you convert my last patches (v3) too, or do I have to do it manually?

It would be cool if you would accept or comment my last patches.

Best regards, Florian

Shawn O. Pearce wrote:

Florian Kberle <FloriansKarten@web.de> wrote:

I'll convert them for you.  Its a simple script and it can be
run multiple times without affecting already converted files.
You could also run it yourself as a final series-end commit by
taking the script out of my commit message in my "bsd" branch.
 

Yea, sorry I have been slacking on that since you posted them.
I still have them in my inbox for me to go through but I've been
quite busy the past few days and haven't had the time needed to
really give them the good reading they deserve.  I'll try to do
it soon.  Probably this weekend.
 

Hi Shawn,

I looked through the licensing info and it is inconsistent. In the feature.xml copyright section you dropped the mention of the GPL while the full license itself was retained later in the section. The latter is correct as we have do have the GPL code that you wrote in the visualization part .

Either we reintroduce mentioning GPL at the start or (you) relicense the graph parts.

To what extent I still have code in there, I have no objections to changing it to either EPL (or LGPL for consistency reasons). I can do the editorial change the license if you confirm that there are no issues.

$ git ls-files -z |xargs -0 grep -l "GNU G"
org.spearce.egit-feature/feature.xml
	inconsistent as mentioned above

org.spearce.egit.core/COPYING
	only mentions LGPL
org.spearce.egit.core/src/org/spearce/egit/core/ResourceList.java
org.spearce.egit.core/src/org/spearce/egit/core/internal/storage/KidCommit.java
org.spearce.egit.core/src/org/spearce/egit/core/internal/storage/KidCommitList.java
org.spearce.egit.core/src/org/spearce/egit/core/internal/storage/KidWalk.java
org.spearce.egit.core/src/org/spearce/egit/core/op/CloneOperation.java
	GPL licensed
org.spearce.egit.ui/COPYING
	only mentions LGPL
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/CommitFileDiffViewer.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/CommitGraphTable.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/CommitMessageViewer.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/CommitNavigationListener.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/FileDiff.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/FileDiffContentProvider.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/FileDiffLabelProvider.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/GenerateHistoryJob.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/GitHistoryPage.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/GraphContentProvider.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/GraphLabelProvider.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/SWTCommit.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/SWTCommitList.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/SWTPlotRenderer.java
org.spearce.egit.ui/src/org/spearce/egit/ui/internal/history/SWTWalk.java
	GPL licensed

-- robin

Robin Rosenberg <robin.rosenberg@dewire.com> wrote:

Oy.  I thought I was correcting it, but I forgot code that I
wrote myself.  Not good.  Very bad mistake.
 

It is my intent to relicense any code I have written in jgit as 3-clause BSD and any code I have written in egit as EPL. The EPL relicense I didn't start the discussion thread for yet, but its where I am trying to go.

EPL for egit code please.  It trivially makes us compatible with the
Eclipse foundation's own code and simplifies our project proposal
with them.
 

Any of this that I wrote, EPL please.

Any of this that I wrote, EPL please. I'm fairly certain Roger wrote some of this code as well, and its derived from your original plotting code. We should make sure it is acceptable to him as well.

Thanks for noticing Robin.

Hi, all

I've got confirmation from all needed parties and pushed the license changes to master. See http://repo.or.cz/w/egit.git?a=commit;h=2baa6eb54706926f74d8ca5a10c7f0448acb5fe6 and http://repo.or.cz/w/egit.git?a=commit;h=53a2cc3f6144ddcc10954d3abf68a5b90ed95248

The changes plus some fixes to the fetch code has been merged and pushed to master.

Philippe, It would be nice if you could check to see that the licensing looks ok. We can hand over the complete e-mail messages separately from the commit comments.

I think we need to complement the UI for fetch plus and couple other fixes and do more testing befire bumping the revision number.

-- robin

Robin: I can check the commit message. If you have a seprate archive of the meails that woudl help too .