On Mon, Mar 21, 2016 at 10:03:12PM +0100, Jan Pokorný wrote: > On 18/03/16 16:16 +0100, Lars Ellenberg wrote: > > On Thu, Mar 17, 2016 at 07:12:19PM +0100, Dejan Muhamedagic wrote: > >> Hi Jan, > >> > >> On Thu, Mar 17, 2016 at 06:47:37PM +0100, Jan Pokorný wrote: > >>> Hello all, > >>> > >>> during latest reviews of packages building on core cluster > >>> infrastructure, it turned out there is a frequented (viral?) issue > >>> with source files declaring unusual licence: GPLv2.1+. > >> > >> Yes, I'd say that it's all coming from a single source. I suspect > >> that nobody's looking at the license, just copies another source > >> file from the same project. Anyway, that's what I did in booth. > >> > >> Who created the first file with this non-existent license is > >> anybody's guess. It could probably be traced, but I doubt that > >> it'd help in any way. > > > > Actually it might. > > > > I think that what happened was this: > > Lars, many thanks for providing the following historical context. > It seems to provide enough background to guide the interpretation > of dubious license notices towards the original intentions. > > Let me premise that from the new perspective, it might have been > a bit preposterous trying to contact every and each (potential) > licensor of the affected files (got a couple of bounces[*], anyway), > just followed the transparency principle, turning them to stand-by > should any coordination be required... > I am sorry if you were looped in without solicitation, especially if > neither project is of your interest anymore. > > > in the early days of heartbeat, way back when, > > source code got "batch tagged" with the license statement: > > http://hg.linux-ha.org/heartbeat-STABLE_3_0/rev/4a67fde00b0b#l1.10 > > 2000/07/26 05:17:18 > > > > Most stuff got tagged with the LGPL 2.1. > > > > Some time later, someone noticed that in some cases, > > a "program" is not a "library", and tried to re-tag > > e.g. "api_test.c" with the GPL 2, > > but without properly taking the actual suggested GPL 2 stanza, > > but by simply dropping "Lesser" and changing "library" to "software". > > http://hg.linux-ha.org/heartbeat-STABLE_3_0/rev/bc508513c4dc#l2.10 > > 2000/08/31 05:23:36 > > > > :-( > > I was made aware of this is downright supported by LGPL 2.1 provision: > > 3. You may opt to apply the terms of the ordinary GNU General Public > License instead of this License to a given copy of the Library. To do > this, you must alter all the notices that refer to this License, so > that they refer to the ordinary GNU General Public License, version > 2, instead of to this License. (If a newer version than version 2 of > the ordinary GNU General Public License has appeared, then you can > specify that version instead if you wish.) Do not make any other > change in these notices. > > Once this change is made in a given copy, it is irreversible for that > copy, so the ordinary GNU General Public License applies to all > subsequent copies and derivative works made from that copy. > > This option is useful when you wish to copy part of the code of the > Library into a program that is not a library. > > So one might think this is just it's imperfect execution. > > > Both changes predate the GPLv3 by seven years. > > Hence the timeless, causation-ignorant, plain logic implication of GPLv3+ > cannot simply hold under the demonstrated reality-check. > > > From there it propagated to ipfail.c and attrd.c, which both became > > *the* template files to start from when writing daemons and extensions > > using the API. > > > > Developers quickly browse their "template", > > their "auto-correct" filter reads "GPL 2", > > which they are content with, > > and in good faith they hack away. > > > > I think it is safe to assume that any developer copying from there meant > > to "stay in project" regarding the licensing. > > Agreed on good will/no surprises being presumed in free software > projects, leading to code being anything that really matters in such > circles. Licensing (and other) paper work is frequently played down > to cargo cult minimum[**]. Unfortunately, a minor inaccuracy there > can go 15+ years down the road unnoticed, as we can observe right now :-) > > > So I move to change it to GPLv2+, for everything that is a "program", > > and LGPLv2.1 for everything that may be viewed as a library. > > > > At least that's how I will correct the wording in the > > affected files in the heartbeat mercurial. > > In the light of the presented historic excursion, that feels natural. > > Assuming no licensors want to speak up, the question now stands: > Is it the same conclusion that has been reached by booth and sbd > package maintainers (Dejan and Andrew respectively, if I follow what's > authoritative nowadays properly) and are these willing to act on it to > prevent the mentioned ambiguous interpretation once forever?
Yes, that's all fine with me. > I will be happy to provide actual patches, Even better :) > but that's not a decision > for me to take. Thanks, Dejan > > In fact, Andrew has already expressed his inclination: > http://oss.clusterlabs.org/pipermail/developers/2016-March/000182.html > > > Thanks for taking this seriously. > > > [*] see the difference of CC field between original and this message > (if you know alternative contacts and think they might be > interested, please point them this thread out) > [**] see also > http://oss.clusterlabs.org/pipermail/users/2016-January/002104.html > > -- > Jan (Poki) _______________________________________________ Developers mailing list Developers@clusterlabs.org http://clusterlabs.org/mailman/listinfo/developers