* clarification on -only and -or-later @ 2019-05-20 18:40 J Lovejoy 2019-05-20 18:52 ` Greg KH 2019-05-21 17:24 ` Bradley M. Kuhn 0 siblings, 2 replies; 20+ messages in thread From: J Lovejoy @ 2019-05-20 18:40 UTC (permalink / raw) To: linux-spdx Hi all, We discussed some example of “interpreting” GPL non-standard notices in terms of when -only or -or-later. I wanted to make sure we all agree. Here are some examples for illustration and rationale: 1) where no version is indicated, the license text of GPL (all versions) tells us what to do, " If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.” - thus, use: SPDX-License-Identifier: GPL-1.0-or-later example: * May be copied or modified under the terms of the GNU General Public License 2) where a version is indicated, but no language to the effect of “or any later version” is included. The license text of GPL (all versions) arguably speaks to this with, "If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation.” Also, the standard header either includes "or (at your option) any later version” or simply removes that to indicate ‘only this version’ - thus, use: SPDX-License-Identifier: GPL-2.0-only examples: * This driver is released to the public under the terms of the * GNU GENERAL PUBLIC LICENSE version 2 * See the file COPYING for details. * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 as published by * the Free Software Foundation. 3) where the license notice in the file simply points to the COPYING file or some other license file that contains the full text of GPL-2.0 This is a tougher call, as there isn’t really any arguably clear call, but my thinking is that we’d use: SPDX-License-Identifier: GPL-2.0-or-later I guess I’d justify that (loosely) because the default license - in this case, GPL-2.0 - contains the default standard header which has the “or any later version” language and no one did anything to remove that or otherwise indicate a limitation, but at least did provide a version of the license, so I’d feel comfortable saying it’s GPL-2.0-or-later. Note: we had long discussions on this kind of example on SPDX and there were some good arguments made to say it could just be GPL-2.0 - so I could be swayed here. At the end of the day, we are stuck with someone who didn’t take the care enough to tell us or use the standard header. And while we could go back to the copyright holder, that may not always be feasible. examples: * This file is subject to the terms and conditions of the GNU General Public * License. See the file "COPYING" in the main directory of this archive * for more details. * See LICENSE.qla2xxx for copyright and licensing details. (where that file is a copy of GPL-2.0) Thanks, Jilayne ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 18:40 clarification on -only and -or-later J Lovejoy @ 2019-05-20 18:52 ` Greg KH 2019-05-20 19:26 ` J Lovejoy 2019-05-21 17:24 ` Bradley M. Kuhn 1 sibling, 1 reply; 20+ messages in thread From: Greg KH @ 2019-05-20 18:52 UTC (permalink / raw) To: J Lovejoy; +Cc: linux-spdx On Mon, May 20, 2019 at 12:40:45PM -0600, J Lovejoy wrote: > Hi all, > > We discussed some example of “interpreting” GPL non-standard notices > in terms of when -only or -or-later. I wanted to make sure we all > agree. Here are some examples for illustration and rationale: This all pertains to the kernel, correct? Other projects may have different viewpoints. > 1) where no version is indicated, the license text of GPL (all > versions) tells us what to do, " If the Program does not specify a > version number of this License, you may choose any version ever > published by the Free Software Foundation.” > - thus, use: SPDX-License-Identifier: GPL-1.0-or-later > > example: > > * May be copied or modified under the terms of the GNU General Public License > > 2) where a version is indicated, but no language to the effect of “or > any later version” is included. The license text of GPL (all versions) > arguably speaks to this with, "If the Program specifies a version > number of this License which applies to it and "any later version", > you have the option of following the terms and conditions either of > that version or of any later version published by the Free Software > Foundation.” > Also, the standard header either includes "or (at your option) any later version” or simply removes that to indicate ‘only this version’ > - thus, use: SPDX-License-Identifier: GPL-2.0-only Or the older format: GPL-2.0 > examples: > > * This driver is released to the public under the terms of the > * GNU GENERAL PUBLIC LICENSE version 2 > * See the file COPYING for details. > > * This program is free software; you can redistribute it and/or modify it > * under the terms of the GNU General Public License version 2 as published by > * the Free Software Foundation. > > 3) where the license notice in the file simply points to the COPYING file or some other license file that contains the full text of GPL-2.0 > This is a tougher call, as there isn’t really any arguably clear call, but my thinking is that we’d use: > SPDX-License-Identifier: GPL-2.0-or-later No, not at all, because the COPYING file is 2.0, why would we assume that any new version would be acceptable? When files have said "see the COPYING file" or text like that, we are marking them: GPL-2.0 > I guess I’d justify that (loosely) because the default license - in > this case, GPL-2.0 - contains the default standard header which has > the “or any later version” language No, there is no such "standard header here". > and no one did anything to remove that or otherwise indicate a > limitation, but at least did provide a version of the license, so I’d > feel comfortable saying it’s GPL-2.0-or-later. Nope, if it does not say "or later" in the text in the file, we will not mark it "or later". That is how we have been operating with these tags for over a year now, we can not change this at this late date, sorry, this has already been decided. > Note: we had long discussions on this kind of example on SPDX and > there were some good arguments made to say it could just be GPL-2.0 - > so I could be swayed here. At the end of the day, we are stuck with > someone who didn’t take the care enough to tell us or use the standard > header. And while we could go back to the copyright holder, that may > not always be feasible. > > examples: > > * This file is subject to the terms and conditions of the GNU General Public > * License. See the file "COPYING" in the main directory of this archive > * for more details. > > * See LICENSE.qla2xxx for copyright and licensing details. > (where that file is a copy of GPL-2.0) Again, "GPL-2.0". Pretty simple :) thanks, greg k-h ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 18:52 ` Greg KH @ 2019-05-20 19:26 ` J Lovejoy 2019-05-20 21:35 ` Allison Randal 0 siblings, 1 reply; 20+ messages in thread From: J Lovejoy @ 2019-05-20 19:26 UTC (permalink / raw) To: Greg KH; +Cc: linux-spdx > On May 20, 2019, at 12:52 PM, Greg KH <gregkh@linuxfoundation.org> wrote: > > On Mon, May 20, 2019 at 12:40:45PM -0600, J Lovejoy wrote: >> Hi all, >> >> We discussed some example of “interpreting” GPL non-standard notices >> in terms of when -only or -or-later. I wanted to make sure we all >> agree. Here are some examples for illustration and rationale: > > This all pertains to the kernel, correct? Other projects may have > different viewpoints. yes, that’s why I was posting on this list :) should have mentioned above this was me trying to remember the discussion from a session at the FSFE event in Barcelona where we discussed some of the “non-standard” license headers. > >> 1) where no version is indicated, the license text of GPL (all >> versions) tells us what to do, " If the Program does not specify a >> version number of this License, you may choose any version ever >> published by the Free Software Foundation.” >> - thus, use: SPDX-License-Identifier: GPL-1.0-or-later >> >> example: >> >> * May be copied or modified under the terms of the GNU General Public License so, I’m assuming you agree with scenario 1? >> >> 2) where a version is indicated, but no language to the effect of “or >> any later version” is included. The license text of GPL (all versions) >> arguably speaks to this with, "If the Program specifies a version >> number of this License which applies to it and "any later version", >> you have the option of following the terms and conditions either of >> that version or of any later version published by the Free Software >> Foundation.” >> Also, the standard header either includes "or (at your option) any later version” or simply removes that to indicate ‘only this version’ >> - thus, use: SPDX-License-Identifier: GPL-2.0-only > > Or the older format: > GPL-2.0 > >> examples: >> >> * This driver is released to the public under the terms of the >> * GNU GENERAL PUBLIC LICENSE version 2 >> * See the file COPYING for details. >> >> * This program is free software; you can redistribute it and/or modify it >> * under the terms of the GNU General Public License version 2 as published by >> * the Free Software Foundation. >> >> 3) where the license notice in the file simply points to the COPYING file or some other license file that contains the full text of GPL-2.0 >> This is a tougher call, as there isn’t really any arguably clear call, but my thinking is that we’d use: >> SPDX-License-Identifier: GPL-2.0-or-later > > No, not at all, because the COPYING file is 2.0, why would we assume > that any new version would be acceptable? > > When files have said "see the COPYING file" or text like that, we are > marking them: > GPL-2.0 > >> I guess I’d justify that (loosely) because the default license - in >> this case, GPL-2.0 - contains the default standard header which has >> the “or any later version” language > > No, there is no such "standard header here". > >> and no one did anything to remove that or otherwise indicate a >> limitation, but at least did provide a version of the license, so I’d >> feel comfortable saying it’s GPL-2.0-or-later. > > Nope, if it does not say "or later" in the text in the file, we will not > mark it "or later”. well, the standard license text includes the exhibit on how to apply the license terms, which includes the recommended standard license notice and that does include "or (at your option) any later version.” - sorry, I wasn’t clear on what I meant above. > > That is how we have been operating with these tags for over a year now, > we can not change this at this late date, sorry, this has already been > decided. This is the one I couldn’t remember how it came out as it can be argued either way, so thanks for clarifying. So, direct reference to a license text file that contains GPL-2.0 will thus be marked as: GPL-2.0-only (or old SPDX identifier, SPDX-2.0) > >> Note: we had long discussions on this kind of example on SPDX and >> there were some good arguments made to say it could just be GPL-2.0 - >> so I could be swayed here. I should have added that I think you can make a viable argument either way. During the discussions with FSF regarding changing the SPDX identifiers for greater clarity, we didn’t get a “default position” of intention from the FSF on what this means. Given the kernel is generally GPL-2.0-only, the position here that reference to the license file is for ’that version only’ makes sense. Just wanted to make sure I remembered and it is all consistent. (I did not meant to imply I wanted to re-open the debate, had enough of that one on the SPDX discussions! ;) >> At the end of the day, we are stuck with >> someone who didn’t take the care enough to tell us or use the standard >> header. And while we could go back to the copyright holder, that may >> not always be feasible. >> >> examples: >> >> * This file is subject to the terms and conditions of the GNU General Public >> * License. See the file "COPYING" in the main directory of this archive >> * for more details. >> >> * See LICENSE.qla2xxx for copyright and licensing details. >> (where that file is a copy of GPL-2.0) > > Again, "GPL-2.0". Pretty simple :) > > thanks, > > greg k-h ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 19:26 ` J Lovejoy @ 2019-05-20 21:35 ` Allison Randal 2019-05-20 22:09 ` J Lovejoy 0 siblings, 1 reply; 20+ messages in thread From: Allison Randal @ 2019-05-20 21:35 UTC (permalink / raw) To: J Lovejoy; +Cc: linux-spdx On 5/20/19 8:26 PM, J Lovejoy wrote: > > Given the kernel is generally GPL-2.0-only, the position here that reference to the license file is for ’that version only’ makes sense. > > Just wanted to make sure I remembered and it is all consistent. (I did not meant to imply I wanted to re-open the debate, had enough of that one on the SPDX discussions! ;) Nod, what I recall from the session in Barcelona was: - If the notice gives a specific license version number, we only use -or-later in the SPDX identifier if the notice explicitly says "or later". The default is "-only" (or just the version number). - If the notice gives no version number, the standard interpretation is that the user can take that as any version of the GPL, which technically would be 1.0-or-later. But, it would also be fine to take the file as 2.0-or-later, which was generally legally preferred. It was not preferred to use 2.0-only in the case of no version number, because it's very possible that existing users are already using the standard interpretation to take those files as GPLv3, and we don't want to cause them problems. (On the flip side, it's massively unlikely that anyone is using the standard interpretation to take those files as GPLv1, so we can safely drop it.) Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 21:35 ` Allison Randal @ 2019-05-20 22:09 ` J Lovejoy 2019-05-20 22:19 ` Allison Randal 0 siblings, 1 reply; 20+ messages in thread From: J Lovejoy @ 2019-05-20 22:09 UTC (permalink / raw) To: Allison Randal; +Cc: linux-spdx > On May 20, 2019, at 3:35 PM, Allison Randal <allison@lohutok.net> wrote: > > On 5/20/19 8:26 PM, J Lovejoy wrote: >> >> Given the kernel is generally GPL-2.0-only, the position here that reference to the license file is for ’that version only’ makes sense. >> >> Just wanted to make sure I remembered and it is all consistent. (I did not meant to imply I wanted to re-open the debate, had enough of that one on the SPDX discussions! ;) > > Nod, what I recall from the session in Barcelona was: > > - If the notice gives a specific license version number, we only use > -or-later in the SPDX identifier if the notice explicitly says "or > later". The default is "-only" (or just the version number). > > - If the notice gives no version number, the standard interpretation is > that the user can take that as any version of the GPL, which technically > would be 1.0-or-later. But, it would also be fine to take the file as > 2.0-or-later, which was generally legally preferred. It was not > preferred to use 2.0-only in the case of no version number, because it's > very possible that existing users are already using the standard > interpretation to take those files as GPLv3, and we don't want to cause > them problems. (On the flip side, it's massively unlikely that anyone is > using the standard interpretation to take those files as GPLv1, so we > can safely drop it.) Considering we have no caselaw on what is “legally preferred” (which is otherwise a vague term we probably ought to avoid :) - and that the GPL text itself states: "If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.” Then, where you have a license notice in file such as, "May be copied or modified under the terms of the GNU General Public License" —> SPDX-License-Identifier: GPL-1.0-or-later While I agree that practically speaking, most people are probably not going to go back to GPL-1.0 - I don’t think we should deviate from what the license explicitly says, i.e., “you may choose any version ever published…” that clearly includes GPL-1.0. From the lengthy conversations about this kind of thing on SPDX legal team calls, this conclusion was pretty non-controversial/everyone agreed. So, I don’t think we should do something different here for any reason. It only muddles things. :) Jilayne ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 22:09 ` J Lovejoy @ 2019-05-20 22:19 ` Allison Randal 2019-05-20 22:52 ` J Lovejoy 0 siblings, 1 reply; 20+ messages in thread From: Allison Randal @ 2019-05-20 22:19 UTC (permalink / raw) To: J Lovejoy; +Cc: linux-spdx On 5/20/19 11:09 PM, J Lovejoy wrote: >> On May 20, 2019, at 3:35 PM, Allison Randal <allison@lohutok.net> wrote: >> >> - If the notice gives no version number, the standard interpretation is >> that the user can take that as any version of the GPL, which technically >> would be 1.0-or-later. But, it would also be fine to take the file as >> 2.0-or-later, which was generally legally preferred. It was not >> preferred to use 2.0-only in the case of no version number, because it's >> very possible that existing users are already using the standard >> interpretation to take those files as GPLv3, and we don't want to cause >> them problems. (On the flip side, it's massively unlikely that anyone is >> using the standard interpretation to take those files as GPLv1, so we >> can safely drop it.) > > Considering we have no caselaw on what is “legally preferred” (which is otherwise a vague term we probably ought to avoid :) - and that the GPL text itself states: Sorry for the vagueness, I was dancing around Chatham House Rule. What I meant was that specific people in the room (no names, but ask me off-list if you don't remember and want to know) said that we should drop GPL 1.0, because "choose any version ever published" allows us to choose to include 1.0 or not, and because there are certain ambiguities in the text of 1.0 that make it worthwhile to drop it. > "If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.” > > Then, where you have a license notice in file such as, > > "May be copied or modified under the terms of the GNU General Public License" > > —> SPDX-License-Identifier: GPL-1.0-or-later > > While I agree that practically speaking, most people are probably not going to go back to GPL-1.0 - I don’t think we should deviate from what the license explicitly says, i.e., “you may choose any version ever published…” that clearly includes GPL-1.0. > > From the lengthy conversations about this kind of thing on SPDX legal team calls, this conclusion was pretty non-controversial/everyone agreed. > > So, I don’t think we should do something different here for any reason. It only muddles things. No one was claiming that an unversioned GPL universally means 2.0-or-later, only that it means we have the option to choose versions. And since we have the option to choose, we should choose 2.0-or-later. Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 22:19 ` Allison Randal @ 2019-05-20 22:52 ` J Lovejoy 2019-05-20 23:15 ` Allison Randal 0 siblings, 1 reply; 20+ messages in thread From: J Lovejoy @ 2019-05-20 22:52 UTC (permalink / raw) To: Allison Randal; +Cc: linux-spdx > On May 20, 2019, at 4:19 PM, Allison Randal <allison@lohutok.net> wrote: > > On 5/20/19 11:09 PM, J Lovejoy wrote: >>> On May 20, 2019, at 3:35 PM, Allison Randal <allison@lohutok.net> wrote: >>> >>> - If the notice gives no version number, the standard interpretation is >>> that the user can take that as any version of the GPL, which technically >>> would be 1.0-or-later. But, it would also be fine to take the file as >>> 2.0-or-later, which was generally legally preferred. It was not >>> preferred to use 2.0-only in the case of no version number, because it's >>> very possible that existing users are already using the standard >>> interpretation to take those files as GPLv3, and we don't want to cause >>> them problems. (On the flip side, it's massively unlikely that anyone is >>> using the standard interpretation to take those files as GPLv1, so we >>> can safely drop it.) >> >> Considering we have no caselaw on what is “legally preferred” (which is otherwise a vague term we probably ought to avoid :) - and that the GPL text itself states: > > Sorry for the vagueness, I was dancing around Chatham House Rule. What I > meant was that specific people in the room (no names, but ask me > off-list if you don't remember and want to know) said that we should > drop GPL 1.0, because "choose any version ever published" allows us to > choose to include 1.0 or not, and because there are certain ambiguities > in the text of 1.0 that make it worthwhile to drop it. ah, right - that was a preference that was expressed. > > >> "If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.” >> >> Then, where you have a license notice in file such as, >> >> "May be copied or modified under the terms of the GNU General Public License" >> >> —> SPDX-License-Identifier: GPL-1.0-or-later >> >> While I agree that practically speaking, most people are probably not going to go back to GPL-1.0 - I don’t think we should deviate from what the license explicitly says, i.e., “you may choose any version ever published…” that clearly includes GPL-1.0. >> >> From the lengthy conversations about this kind of thing on SPDX legal team calls, this conclusion was pretty non-controversial/everyone agreed. >> >> So, I don’t think we should do something different here for any reason. It only muddles things. > > No one was claiming that an unversioned GPL universally means > 2.0-or-later, only that it means we have the option to choose versions. > And since we have the option to choose, we should choose 2.0-or-later. yeah, one can make that viable argument, I guess I’d just prefer that we don’t put ourselves in a position to need to explain too much. Preferably the SPDX identifier is an obvious expression of what the license notice stated or the copyright holders cleaned up any ambiguities. > > Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 22:52 ` J Lovejoy @ 2019-05-20 23:15 ` Allison Randal 0 siblings, 0 replies; 20+ messages in thread From: Allison Randal @ 2019-05-20 23:15 UTC (permalink / raw) To: J Lovejoy; +Cc: linux-spdx On 5/20/19 11:52 PM, J Lovejoy wrote: >> On May 20, 2019, at 4:19 PM, Allison Randal <allison@lohutok.net> wrote: >> >> No one was claiming that an unversioned GPL universally means >> 2.0-or-later, only that it means we have the option to choose versions. >> And since we have the option to choose, we should choose 2.0-or-later. > > yeah, one can make that viable argument, I guess I’d just prefer that we don’t put ourselves in a position to need to explain too much. Preferably the SPDX identifier is an obvious expression of what the license notice stated or the copyright holders cleaned up any ambiguities. I don't personally have a strong opinion either way. I appreciate the legal consistency of choosing 2.0 where we can (consistent with 2.0 in LICENSES/preferred, and 1.0 in LICENSES/deprecated), and I respect the people who were arguing for dropping 1.0. So, if I had to make the call, I'd probably go with 2.0-or-later. But, I'm fine with whatever is chosen. Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-20 18:40 clarification on -only and -or-later J Lovejoy 2019-05-20 18:52 ` Greg KH @ 2019-05-21 17:24 ` Bradley M. Kuhn 2019-05-21 18:05 ` J Lovejoy [not found] ` <5EB6B416-F24C-4741-BC0E-6C1896E7A705@jilayne.com> 1 sibling, 2 replies; 20+ messages in thread From: Bradley M. Kuhn @ 2019-05-21 17:24 UTC (permalink / raw) To: linux-spdx J Lovejoy wrote: > 3) where the license notice in the file simply points to the COPYING file or some other license file that contains the full text of GPL-2.0 > This is a tougher call, as there isn’t really any arguably clear call, but > my thinking is that we’d use: > SPDX-License-Identifier: GPL-2.0-or-later I agree that one can use GPL-1.0-or-later in this case well (which was discussed down thread), but I also agree with the argument (also downthread) that there is no *requirement* to include GPL-1.0 in the mix. The text of the COPYING file (i.e., GPLv2) is clear on this point, if we have code that does "not specify a version number of this License, you may choose any version ever published by the Free Software Foundation." Jilayne and I did a pretty deep dive on this question of the 'no version number specified' and I think our discussions made us sure that it does *not* mean GPL-2.0-only, because of the text above. I checked with Fontana too and he agrees with this as well. Meta note: I've got a hectic week so I am not available to look at any Thomas' patch sets (and the threads they're generating) until this weekend, but I've set aside time on this Sunday morning for it. Looking forward to it! -- Bradley M. Kuhn Pls. support the charity where I work, Software Freedom Conservancy: https://sfconservancy.org/supporter/ ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-21 17:24 ` Bradley M. Kuhn @ 2019-05-21 18:05 ` J Lovejoy 2019-05-22 13:23 ` Greg KH [not found] ` <5EB6B416-F24C-4741-BC0E-6C1896E7A705@jilayne.com> 1 sibling, 1 reply; 20+ messages in thread From: J Lovejoy @ 2019-05-21 18:05 UTC (permalink / raw) To: Bradley M. Kuhn; +Cc: linux-spdx HI Bradley, Thanks for weighing in there. I think my original examples got a but lost in the various back and forth. So, let me reproduce and re-match: 1) where no version is indicated, the license text of GPL (all versions) tells us what to do, " If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.” - thus, use: SPDX-License-Identifier: GPL-1.0-or-later example: * May be copied or modified under the terms of the GNU General Public License This is what Allison and I were going back and forth on. Net sum being: I was pointing out that under a literal reading of the license, such a unclear reference to just “GPL” would be GPL-1.0-or-later I think this is where your point is spot on and confirms my memory of the various discussions: > I agree that one can use GPL-1.0-or-later in this case well (which was > discussed down thread), but I also agree with the argument (also downthread) > that there is no *requirement* to include GPL-1.0 in the mix. The text of > the COPYING file (i.e., GPLv2) is clear on this point, if we have code that > does "not specify a version number of this License, you may choose any > version ever published by the Free Software Foundation." > > Jilayne and I did a pretty deep dive on this question of the 'no version > number specified' and I think our discussions made us sure that it does > *not* mean GPL-2.0-only, because of the text above. I checked with > Fontana too and he agrees with this as well. I think what I was looking for here, was confirmation as to whether we want to do the “literal” GPL-1.0-or-later option that the license provides for, or trigger the option to “choose any version” and go with GPL-2.0-or-later for consistency of v2 across the kernel and for other reasons I believe you raised regarding GPL-1.0 Thoughts? thanks, Jilayne > On May 21, 2019, at 11:24 AM, Bradley M. Kuhn <bkuhn@ebb.org> wrote: > > J Lovejoy wrote: >> 3) where the license notice in the file simply points to the COPYING file or some other license file that contains the full text of GPL-2.0 > >> This is a tougher call, as there isn’t really any arguably clear call, but >> my thinking is that we’d use: > >> SPDX-License-Identifier: GPL-2.0-or-later > > I agree that one can use GPL-1.0-or-later in this case well (which was > discussed down thread), but I also agree with the argument (also downthread) > that there is no *requirement* to include GPL-1.0 in the mix. The text of > the COPYING file (i.e., GPLv2) is clear on this point, if we have code that > does "not specify a version number of this License, you may choose any > version ever published by the Free Software Foundation." > > Jilayne and I did a pretty deep dive on this question of the 'no version > number specified' and I think our discussions made us sure that it does > *not* mean GPL-2.0-only, because of the text above. I checked with > Fontana too and he agrees with this as well. > > Meta note: I've got a hectic week so I am not available to look at > any Thomas' patch sets (and the threads they're generating) until this weekend, > but I've set aside time on this Sunday morning for it. Looking forward to it! > > -- > Bradley M. Kuhn > > Pls. support the charity where I work, Software Freedom Conservancy: > https://sfconservancy.org/supporter/ ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-21 18:05 ` J Lovejoy @ 2019-05-22 13:23 ` Greg KH 2019-05-22 13:53 ` Allison Randal 0 siblings, 1 reply; 20+ messages in thread From: Greg KH @ 2019-05-22 13:23 UTC (permalink / raw) To: J Lovejoy; +Cc: Bradley M. Kuhn, linux-spdx On Tue, May 21, 2019 at 12:05:37PM -0600, J Lovejoy wrote: > > I agree that one can use GPL-1.0-or-later in this case well (which was > > discussed down thread), but I also agree with the argument (also downthread) > > that there is no *requirement* to include GPL-1.0 in the mix. The text of > > the COPYING file (i.e., GPLv2) is clear on this point, if we have code that > > does "not specify a version number of this License, you may choose any > > version ever published by the Free Software Foundation." > > > > Jilayne and I did a pretty deep dive on this question of the 'no version > > number specified' and I think our discussions made us sure that it does > > *not* mean GPL-2.0-only, because of the text above. I checked with > > Fontana too and he agrees with this as well. > > I think what I was looking for here, was confirmation as to whether we > want to do the “literal” GPL-1.0-or-later option that the license > provides for, or trigger the option to “choose any version” and go > with GPL-2.0-or-later for consistency of v2 across the kernel and for > other reasons I believe you raised regarding GPL-1.0 I don't understand. Can you point to any files in the kernel where we have used the "GPL-1.0+" marking incorrectly? thanks, greg k-h ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 13:23 ` Greg KH @ 2019-05-22 13:53 ` Allison Randal 2019-05-22 14:00 ` Greg KH 0 siblings, 1 reply; 20+ messages in thread From: Allison Randal @ 2019-05-22 13:53 UTC (permalink / raw) To: Greg KH, J Lovejoy; +Cc: linux-spdx On 5/22/19 2:23 PM, Greg KH wrote: > On Tue, May 21, 2019 at 12:05:37PM -0600, J Lovejoy wrote: >> >> I think what I was looking for here, was confirmation as to whether we >> want to do the “literal” GPL-1.0-or-later option that the license >> provides for, or trigger the option to “choose any version” and go >> with GPL-2.0-or-later for consistency of v2 across the kernel and for >> other reasons I believe you raised regarding GPL-1.0 > > I don't understand. Can you point to any files in the kernel where we > have used the "GPL-1.0+" marking incorrectly? Jilayne's question wasn't about current usage in the kernel, it was about what we should do in this cleanup process when we get to files where the license notice doesn't have an explicit GPL version number or include the "or later" text. Thomas hasn't gotten to those patterns yet in his batch processing. Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 13:53 ` Allison Randal @ 2019-05-22 14:00 ` Greg KH 2019-05-22 14:20 ` Thomas Gleixner ` (2 more replies) 0 siblings, 3 replies; 20+ messages in thread From: Greg KH @ 2019-05-22 14:00 UTC (permalink / raw) To: Allison Randal; +Cc: J Lovejoy, linux-spdx On Wed, May 22, 2019 at 09:53:48AM -0400, Allison Randal wrote: > On 5/22/19 2:23 PM, Greg KH wrote: > > On Tue, May 21, 2019 at 12:05:37PM -0600, J Lovejoy wrote: > >> > >> I think what I was looking for here, was confirmation as to whether we > >> want to do the “literal” GPL-1.0-or-later option that the license > >> provides for, or trigger the option to “choose any version” and go > >> with GPL-2.0-or-later for consistency of v2 across the kernel and for > >> other reasons I believe you raised regarding GPL-1.0 > > > > I don't understand. Can you point to any files in the kernel where we > > have used the "GPL-1.0+" marking incorrectly? > > Jilayne's question wasn't about current usage in the kernel, it was > about what we should do in this cleanup process when we get to files > where the license notice doesn't have an explicit GPL version number or > include the "or later" text. Thomas hasn't gotten to those patterns yet > in his batch processing. Ah, ok. But note, we have already marked such files as "GPL-1.0+" in the past, so any change in that behavior would require us go back and change what we did, showing the justification for that. I would stick to the rule of what we have already done in these cases, it's simpler and seems to make sense of a crazy situation. thanks, greg k-h ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 14:00 ` Greg KH @ 2019-05-22 14:20 ` Thomas Gleixner 2019-05-22 14:30 ` Allison Randal 2019-05-22 14:22 ` Allison Randal 2019-05-22 15:03 ` J Lovejoy 2 siblings, 1 reply; 20+ messages in thread From: Thomas Gleixner @ 2019-05-22 14:20 UTC (permalink / raw) To: Greg KH; +Cc: Allison Randal, J Lovejoy, linux-spdx [-- Attachment #1: Type: text/plain, Size: 1727 bytes --] On Wed, 22 May 2019, Greg KH wrote: > On Wed, May 22, 2019 at 09:53:48AM -0400, Allison Randal wrote: > > On 5/22/19 2:23 PM, Greg KH wrote: > > > On Tue, May 21, 2019 at 12:05:37PM -0600, J Lovejoy wrote: > > >> > > >> I think what I was looking for here, was confirmation as to whether we > > >> want to do the “literal” GPL-1.0-or-later option that the license > > >> provides for, or trigger the option to “choose any version” and go > > >> with GPL-2.0-or-later for consistency of v2 across the kernel and for > > >> other reasons I believe you raised regarding GPL-1.0 > > > > > > I don't understand. Can you point to any files in the kernel where we > > > have used the "GPL-1.0+" marking incorrectly? > > > > Jilayne's question wasn't about current usage in the kernel, it was > > about what we should do in this cleanup process when we get to files > > where the license notice doesn't have an explicit GPL version number or > > include the "or later" text. Thomas hasn't gotten to those patterns yet > > in his batch processing. > > Ah, ok. > > But note, we have already marked such files as "GPL-1.0+" in the past, > so any change in that behavior would require us go back and change what > we did, showing the justification for that. > > I would stick to the rule of what we have already done in these cases, > it's simpler and seems to make sense of a crazy situation. Yes and no. There is a good reason to remove the GPL1+ crap completely and I already got permission from Redhat to change their GPL/GPL'ed notices to GPL-2.0-or-later. So once we get to that pile we might at least try to talk with the copyright holders and clarify it. Each odd license which gets removed is a win. Thanks, tglx ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 14:20 ` Thomas Gleixner @ 2019-05-22 14:30 ` Allison Randal 2019-05-22 15:45 ` Greg KH 0 siblings, 1 reply; 20+ messages in thread From: Allison Randal @ 2019-05-22 14:30 UTC (permalink / raw) To: Thomas Gleixner; +Cc: linux-spdx On 5/22/19 3:20 PM, Thomas Gleixner wrote: > Yes and no. There is a good reason to remove the GPL1+ crap completely and > I already got permission from Redhat to change their GPL/GPL'ed notices to > GPL-2.0-or-later. So once we get to that pile we might at least try to talk > with the copyright holders and clarify it. Each odd license which gets > removed is a win. We don't necessarily have to do everything in one pass. We could keep this round of cleanup strictly to an exact substitution of GPL-1.0-or-later, and do a separate pass at some point in the future to deprecate and remove GPL-1.0 where possible. I'm not sure whether it would be easier to do it all now or save the deprecation for later. I guess we can see how many files match the patterns with no version numbers. (With 5 files it's easy to contact the copyright holders, hundreds would be difficult.) Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 14:30 ` Allison Randal @ 2019-05-22 15:45 ` Greg KH 2019-05-22 19:04 ` Bradley M. Kuhn 0 siblings, 1 reply; 20+ messages in thread From: Greg KH @ 2019-05-22 15:45 UTC (permalink / raw) To: Allison Randal; +Cc: Thomas Gleixner, linux-spdx On Wed, May 22, 2019 at 10:30:07AM -0400, Allison Randal wrote: > On 5/22/19 3:20 PM, Thomas Gleixner wrote: > > Yes and no. There is a good reason to remove the GPL1+ crap completely and > > I already got permission from Redhat to change their GPL/GPL'ed notices to > > GPL-2.0-or-later. So once we get to that pile we might at least try to talk > > with the copyright holders and clarify it. Each odd license which gets > > removed is a win. If we get permission, that's great! But given that we already have 53 files with that tag today, the odds of someone contacting all of those owners is going to be hard :( > We don't necessarily have to do everything in one pass. We could keep > this round of cleanup strictly to an exact substitution of > GPL-1.0-or-later, and do a separate pass at some point in the future to > deprecate and remove GPL-1.0 where possible. I would be amazed if anyone ever does a "follow-on" pass of the tree after we finish with these, other than to add tags to files that sneak in without them :) thanks, greg k-h ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 15:45 ` Greg KH @ 2019-05-22 19:04 ` Bradley M. Kuhn 0 siblings, 0 replies; 20+ messages in thread From: Bradley M. Kuhn @ 2019-05-22 19:04 UTC (permalink / raw) To: linux-spdx We should take care not to conflate changes made because: (a) all relevant copyright holders consented, vs. (b) consent not available from the copyright holders. (a) and (b) differ greatly, both politically and legally. With (a), we can certainly ask copyright holders for any notice changes that seem useful, expedient, and would help Linux as a project. (e.g., if, as Thomas says, Red Hat is sole copyright holder of some code and agrees to relicense it from GPL-1.0-or-later to GPL-2.0-or-later, I see no reason not to gleefully accept that and make the change (copyright holders assent well-documented in the commit log with relevant Signed-Off-By: , of course). Similarly, (to "cross" the two active threads a bit), that we really have no worries about old-notice-preservation if all relevant copyright holders assents to replace their old notice with an SPDX Identifier. However, with (b), judgment and risk analysis, both of legal and political nature, will always be required. During the "Great SPDX GPL Identifier Change" last year, Jilayne and I dug deep on the legal judgment/risk-analysis about some of these notices -- in collaboration with FSF (as license steward) and many others in the FOSS licensing community -- to come to conclusions about the two ambiguous cases Jilayne listed as (1) and (3) initially in this thread. Jilayne's suggestions for these cases were presumably informed by that, and I suspect that's why she's recommending them as -or-later conclusions. Nevertheless, if there is political risk (or even, as Greg points out, annoying work not worth doing) involved with moving GPL-1.0-or-later code to (say) GPL-2.0-or-later instead, then there is no reason to do so. Furthermore, such a change also relates to this point: There is danger of taking things that legal analysis concludes are '-or-later' and turning them into '-only'. Narrowing from '-or-later' to '-only' is always permitted by downstream (and also the effective license of the whole work of Linux will undoubtedly remain GPL-2.0-only). However, so many people have expressed a desire for accurate, complete, and as-broad-as-legally-possible file-by-file licensing inventory. This project exists, presumably, to serve those requests. We should strive to make sure that, on a "file-by-file level", this project doesn't inadvertently narrow permissions unduly. Furthermore, combining the work of notice-replacement with license *changing* adds undue risk; the two activities should be fully separated by both time and workflow. If the upshot of this *also* means we live with more GPL-1.0-or-later notices floating around, I don't think it's that big of a deal. Better that than annoyed contributors, and, more importantly, downstream users who wish to take a the more liberal license for some code in the Linux tree, etc. As always, IANAL and TINLA, -- Bradley M. Kuhn Pls. support the charity where I work, Software Freedom Conservancy: https://sfconservancy.org/supporter/ ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 14:00 ` Greg KH 2019-05-22 14:20 ` Thomas Gleixner @ 2019-05-22 14:22 ` Allison Randal 2019-05-22 15:03 ` J Lovejoy 2 siblings, 0 replies; 20+ messages in thread From: Allison Randal @ 2019-05-22 14:22 UTC (permalink / raw) To: Greg KH; +Cc: linux-spdx On 5/22/19 3:00 PM, Greg KH wrote: > > But note, we have already marked such files as "GPL-1.0+" in the past, > so any change in that behavior would require us go back and change what > we did, showing the justification for that. > > I would stick to the rule of what we have already done in these cases, > it's simpler and seems to make sense of a crazy situation. Fair. If we have a precedent of keeping GPL-1.0 in the SPDX license notices, then we should be consistent and follow that rule in this cleanup process. Allison ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: clarification on -only and -or-later 2019-05-22 14:00 ` Greg KH 2019-05-22 14:20 ` Thomas Gleixner 2019-05-22 14:22 ` Allison Randal @ 2019-05-22 15:03 ` J Lovejoy 2 siblings, 0 replies; 20+ messages in thread From: J Lovejoy @ 2019-05-22 15:03 UTC (permalink / raw) To: Greg KH; +Cc: Allison Randal, linux-spdx > On May 22, 2019, at 8:00 AM, Greg KH <gregkh@linuxfoundation.org> wrote: > > On Wed, May 22, 2019 at 09:53:48AM -0400, Allison Randal wrote: >> On 5/22/19 2:23 PM, Greg KH wrote: >>> On Tue, May 21, 2019 at 12:05:37PM -0600, J Lovejoy wrote: >>>> >>>> I think what I was looking for here, was confirmation as to whether we >>>> want to do the “literal” GPL-1.0-or-later option that the license >>>> provides for, or trigger the option to “choose any version” and go >>>> with GPL-2.0-or-later for consistency of v2 across the kernel and for >>>> other reasons I believe you raised regarding GPL-1.0 >>> >>> I don't understand. Can you point to any files in the kernel where we >>> have used the "GPL-1.0+" marking incorrectly? >> >> Jilayne's question wasn't about current usage in the kernel, it was >> about what we should do in this cleanup process when we get to files >> where the license notice doesn't have an explicit GPL version number or >> include the "or later" text. Thomas hasn't gotten to those patterns yet >> in his batch processing. > > Ah, ok. > > But note, we have already marked such files as "GPL-1.0+" in the past, > so any change in that behavior would require us go back and change what > we did, showing the justification for that. > > I would stick to the rule of what we have already done in these cases, > it's simpler and seems to make sense of a crazy situation. > Thanks Allison for helping to clarify and yes, if there is a pattern already, it’d be really helpful for someone to send me a few examples. The examples you’ve seen already may not match some of the “messy files” I’m looking at, but it’d still be helpful. thanks, Jilayne ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <5EB6B416-F24C-4741-BC0E-6C1896E7A705@jilayne.com>]
* Re: clarification on -only and -or-later [not found] ` <5EB6B416-F24C-4741-BC0E-6C1896E7A705@jilayne.com> @ 2019-05-21 21:14 ` Bradley M. Kuhn 0 siblings, 0 replies; 20+ messages in thread From: Bradley M. Kuhn @ 2019-05-21 21:14 UTC (permalink / raw) To: linux-spdx Jilayne, thanks for noting that discussion about (1) and (3) got mixed up downthread. OTOH, the situations are quite similar anyway. To clarify, I was indeed commenting on your original item (3), not (1): Jilayne wrote at the start of the thread: > 3) where the license notice in the file simply points to the COPYING file or > some other license file that contains the full text of GPL-2.0 > his is a tougher call, as there isn’t really any arguably clear call, but my > thinking is that we’d use: > SPDX-License-Identifier: GPL-2.0-or-later I think that solution you propose above is correct on item (3), as discussed in my previous email. Since you asked me in your follow up to also comment on your original item (1), that's below: J Lovejoy wrote: > 1) where no version is indicated, the license text of GPL (all versions) > tells us what to do, " If the Program does not specify a version number > of this License, you may choose any version ever published by the Free > Software Foundation.” > - thus, use: SPDX-License-Identifier: GPL-1.0-or-later > example: > * May be copied or modified under the terms of the GNU General Public > License The situation doesn't differ much from (3). In both cases, a reference is made to "GPL" without specifying a version number. GPL (all versions) say that if it "does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.". This is the very issue that you and I discussed at length when FSF was lobbying for identifier changes, and we sorted out that indeed that, in any situation where a version number isn't specified, the quoted text causes maximal options. So, in both cases (1) and (3), downstream licensors (which the Linux project is in this scenario, since it received the contribution 'from someone') can chose the whole spectrum they want, including GPL-1.0-or-later. I think choosing the "-only" versions -- while permissible as downstream licensor (i.e., you can always opt yourself to narrow an -or-later into an -only) -- unnecessarily limits possibility for code sharing. Linux is clearly committed to keeping file-by-file inventory of licensing information, and given that commitment, the project should strive to keep the most permissive version of the license it can. (This situation is akin to the dual-licensed: "BSD-3-Clause OR GPL-2.0-or-later" -- while BSD-3-Clause can easily be dropped, it's not recommended for various reasons.) However, as we discussed in Barcelona, the code sharing argument for GPL-1.0 inclusion is minimal. No one could think of *any* project that still *intentionally* includes GPL-1.0-only and/or GPL-1.0-or-later in its license choices. If we have an example of that, particular of a program written in C, that would be a good counter example and would lean me toward thinking staying with GPL-1.0-or-later in these cases makes sense. Absent any example like that, GPL-2.0-or-later is the way to go for examples (1) and (3). -- Bradley M. Kuhn Pls. support the charity where I work, Software Freedom Conservancy: https://sfconservancy.org/supporter/ ^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2019-05-22 19:10 UTC | newest] Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2019-05-20 18:40 clarification on -only and -or-later J Lovejoy 2019-05-20 18:52 ` Greg KH 2019-05-20 19:26 ` J Lovejoy 2019-05-20 21:35 ` Allison Randal 2019-05-20 22:09 ` J Lovejoy 2019-05-20 22:19 ` Allison Randal 2019-05-20 22:52 ` J Lovejoy 2019-05-20 23:15 ` Allison Randal 2019-05-21 17:24 ` Bradley M. Kuhn 2019-05-21 18:05 ` J Lovejoy 2019-05-22 13:23 ` Greg KH 2019-05-22 13:53 ` Allison Randal 2019-05-22 14:00 ` Greg KH 2019-05-22 14:20 ` Thomas Gleixner 2019-05-22 14:30 ` Allison Randal 2019-05-22 15:45 ` Greg KH 2019-05-22 19:04 ` Bradley M. Kuhn 2019-05-22 14:22 ` Allison Randal 2019-05-22 15:03 ` J Lovejoy [not found] ` <5EB6B416-F24C-4741-BC0E-6C1896E7A705@jilayne.com> 2019-05-21 21:14 ` Bradley M. Kuhn
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.