J Lovejoy wrote:
(And to give credit where credit is due, Bradley's input during that challenging "negotiation" was very helpful. :)
😊 … thank you!
I'd written today:
So, this problem that Thomas notes above is definitely an error by the SPDX project, *just like* the one that exists for the deprecated “GPL-2.0”
J Lovejoy replied:
To be clear, the GPL-2.0 identifier was never an error by the SPDX team - we were always very clear as to what it meant/means.
… but notwithstanding a clear definition of a moniker (which I agree indeed you've made for most SPDX identifiers), if that definition fails to adequately match historically understanding (and/or fails to take into account nuances in the document it represents), confusion ensues for users. Users *were* confused about “GPL-2.0” (remember, we did a small (admittedly non-scientific) survey at a session at a conference — FOSDEM I think it was?)
Most SPDX *users* won't speak its defined terms fluently; I suspect most of Linux's licensors (and even most licensees) don't speak SPDX fluently, so presumably you want SPDX identifiers to have some intuitiveness — particularly for the use case of linux-spdx, which requires the identifiers to be *both* human-readable and machine-readable.
This is relevant to the copyleft-next-0.3.1 identifier. SPDX could define “copyleft-next-0.3.1” to mean for SPDX purposes: “the text of copyleft-next without any options in its terms exercised/removed” (— although I note https://spdx.org/licenses/copyleft-next-0.3.1.html seems to be wholly silent regarding options exercising/removing). However, there is currently confusion — shown in the fact that Thomas still asked:
If I want to remove this option, then how do I express this with a SPDX license identifier? Sigh!
… upon noticing this part of copyleft-next:
- Unless I explicitly remove the option of Distributing Covered Works
- under Later Versions, You may Distribute Covered Works under any Later
- Version.
Anyway, I'm pointing out SPDX's shortcomings on this point *not* to captiously admonish SPDX, but rather to point out that any issues with SPDX identifiers and their formal definitions shouldn't influence a decision about what licenses are acceptable for inclusion as dual-license options in Linux.
Plus, I remain hopeful that over the long-term, the SPDX project will take feedback from efforts like linux-spdx to solve the kinds of problems that have come up in this thread and others.
Finally, I've already started a sub-thread on the copyleft-next list to start discussing maybe the license (in future versions) shouldn't have this option anyway (for unrelated policy reasons). That might yield a side-benefit of making the problem evaporate entirely for SPDX. (Anyway, after 25 years of living with GPL's “-or-later vs. -only” mess — I, for one, am convinced new licenses like copyleft-next should try very hard to not repeat that mistake.)
-- bkuhn