It is precisely because people like you refuse to consider changes to antiquated policies that I lead these long threads. Don't presume that I have anything more important to do than to convince ASF, in which you are but one member, about things important to me.
My 20+ years of experience in open source law and with open source projects have given me great endurance to overcome license compatibility issues. For example, I led the Apache negotiations that caused GPLv3 to finally accept all ALv2 software. I have repeatedly fought the silly "static linking" doctrine that poisoned GPL software for Larger Works long before I starting participating at ASF; I haven't stopped that legal argument either and would welcome a court decision.
But I assure you that until you get at least one open source attorney publicly to debate this issue on behalf of Sam and Greg and Ross, then I have continued patience. I actually enjoy this opportunity to hone my arguments here in public. And you have a delete key.
Not one single attorney has asked me to shut up.
"If this were legal advice it would have been accompanied by a bill."
This seemingly endless discussion has been going on for so long now that I am beginning to believe Larry just doesn’t care what the impact is on downstream users. Everyone here understands the argument he is making, but he doesn’t seem to understand the reason why the ASF membership prefers the policy the way it is.
Personally, I am tired of these arguments and wish we would all just ignore this thread.
On Aug 1, 2015, at 4:48 PM, David Jencks <firstname.lastname@example.org> wrote:
I hope that never happens and I expect if it does the source code will be removed as soon as anyone notices.
While it seems clear even to me as well as you that when _apache_ publishes such contributed code all the license conditions are satisfied, that’s not the point of the policy. The problem with including such source is that anyone downstream who wants to modify the work and distribute it in binary-only form can’t. At best they have to look at the provenance and history of every modified file to determine what license conditions apply to that file to see whether they have to publish the changes or not.
In the part if IBM I work in, I’d personally get to do this for any such apache project I managed to convince the powers-that-be would be a good idea to use. That would make my life hell, and I’d probably avoid trying to reuse apache software we’d have to modify in proprietary ways. In my experience IBM is usually pretty good about pushing useful changes back, but sometimes adapting code to different circumstances is not useful to contribute back, and sometimes it takes a long time.
Here’s a recent concrete example: IBM’s recently released Liberty WAS now uses the geronimo-yoko orb. We started on this effort about a year ago, and immediately started fixing bugs. It only took a couple months to get legal approval to actually use the orb, but for a variety of reasons we didn’t get legal approval to contribute the bug fixes back until very recently. If we had to have scanned the code for which license each file was under and gotten legal approval to contribute changes back before starting work, we would not have released by now.
As far as I can tell what you are proposing is perfectly legal but would make apache software unusable for many of its current users.