Picking a license
in [Python]
|
Prev: 2to3 issues with execfile on python 3.0 on files with encoding
Next: Extract a bordered, skewed rectangle from an image
From: Paul Boddie on 9 May 2010 17:21 On 9 Mai, 21:07, Patrick Maupin <pmau...(a)gmail.com> wrote: > On May 9, 1:02 pm, Paul Boddie <p...(a)boddie.org.uk> wrote: > > > > People often argue > > that the GPL only cares about the software's freedom, not the > > recipient's freedom, which I find to be a laughable claim because if > > one wanted to point at something the GPL places higher than anything > > else, it would be the "four freedoms" preserved for each user's > > benefit. > > Well, I don't think you saw me arguing it that way. I will say, just > like anything else, that there is a cost associated with using GPL > software, and it is not necessarily a cost that I want to impose on > users of all my software. I didn't say that you personally argued that way, but people do argue that way. In fact, it's understandable that this is how some people attempt to understand the GPL - the software maintains a particular state of openness - but they miss the final step in the reasoning which leads them to see that the licence preserves a set of privileges for recipients as well. The "cost" with the GPL is that people cannot take GPL-licensed software and just do whatever they want with it, although it is also the case that permissive licences also have a set of conditions associated with each of them as well, albeit ones which do not mandate the delivery of the source code to recipients. Thus, the observation of software licences can never be about taking code which was publicly available and combining it without thought to what those licences say. Thus, remarks about Cisco and Linksys - that they were somehow "caught out" - are disingenuous: if you're in the business of distributing software, particularly if that software itself has a restrictive licence, you cannot claim ignorance about licensing or that you just "found some good code". > > Really, copyleft licences are all about treating all recipients of the > > software and modified versions or extensions of the software in the > > same way: that someone receiving the software, in whatever state of > > enhancement, has all the same privileges that the individual or > > organisation providing the software to them enjoyed; > > Sure, and for a major work I think that's great, especially if it > helps attract developers. Sometimes I see people GPL little 100 line > libraries (of often not very good code quality) in a clear attempt to > have the tail wag the dog, and that's laughably pathetic. Why is it pathetic that someone gets to choose the terms under which their work is made available? By default, if I release something without any licence, the recipient has very few privileges with respect to that work: it's literally a case of "all rights reserved" for the creator. And if it's such a trivial library then why not reimplement the solution yourself? > > those "four > > freedoms" should still apply to whatever software they received. That > > this is achieved by asking that everyone make the same commitment to > > end-user freedoms (or privileges), yet is seen as unreasonable or > > actually perceived as coercion by some, says a great deal about the > > perspective of those complaining about it. > > Well, I *do* think it's, maybe not unreasonable, but certainly > unrealistic, for the author of a small library to attempt to leverage > control over several potentially much larger works by placing the > small library under the GPL, so in general I don't do it. I dislike the way that when someone releases something under the GPL, it is claimed that they are coercing or attempting to "leverage" something. They have merely shared something on their terms. If you don't like the terms, don't use their software. > I also > happen to believe that there are a lot of people (perhaps like Carl > Banks if I understand his post correctly) who make money delivering > small customized solutions to sit on top of proprietary software > solutions. If I can save one of these guys some time, perhaps they > will contribute back. If I use the GPL, I will have insured that one > of these guys cannot possibly link my software to, e.g. Oracle, so he > has to reinvent the wheel. So, for some use-cases, I sincerely > believe that the GPL license creates unnecessary, wasteful friction. But it is not universally true that GPL-licensed software cannot be linked to proprietary software: there are a number of caveats in the GPL covering cases where existing proprietary systems are in use. Otherwise, you'd never have GPL-licensed software running on proprietary systems at all. > But the tone of your last statement and some of your statements below > make it abundantly clear that you've made up your mind about my morals > and aren't at all interested in my reasoning. Not at all. Recently, I've had the misfortune to hear lots of arguments about how the GPL supposedly restricts stuff like "collaboration" and "growth" despite copious evidence to the contrary, usually from people who seem to be making a career of shouting down the GPL or the FSF at every available occasion. Now I'm not saying that you have the same apparent motivations as these people, but I maintain that when someone claims that people are "forced" to share their work when they voluntarily make use of someone else's work, or that they are at the peril of some "moral hazard", it does have a lot to say about their perspective. (Not least because people are only obliged to make their work available under a GPL-compatible licence so that people who are using the combined work may redistribute it under the GPL. You yourself have mentioned elsewhere in this discussion one well-known software project that is not GPL-licensed but was effectively distributed under the GPL to most of its users for a considerable period of time.) [...] > > Yes, but what irritates a lot of people is when you see other people > > arguing that some other random person should license their own > > software permissively because it's "better" or "more free" when what > > they really mean is that "I could use it to make a proprietary > > product". > > I'm not telling anybody what to do. I'm just explaining why I usually > use the MIT license for things I write, and will often not consider > using a library licensed under the GPL. What irritated me enough to > comment on this thread was the IMHO sanctimonious and inflammatory > "Unless you place such a low value the freedom of your users". It is hardly a rare occurrence now that I come across someone who has written in some corner of the Internet, "It's a shame project XYZ is GPL-licensed because I can't use it for commercial software development. Can the project maintainers not choose another licence?" Sometimes, someone who is seeking licensing advice might not want to be unpopular and might choose a permissive licence because people reassure them that their project will only be widely used if the licence lets people use it "commercially" (or, in other words, in proprietary software). My impression is that many in the core community around Python seem to emphasise such popularity over all other concerns. What I want to point out, and some have done so much more directly than I have in other forums and in other discussions, is that some advice about licensing often stems from a direct motivation amongst those giving the advice to secure preferential terms for themselves, and that although such advice may be dressed up as doing the "right" or "best" thing, those giving the advice stand to gain directly and even selfishly from having their advice followed. I'm not saying you have done this, but this is frequently seen in the core Python community, such that anyone suggesting a copyleft licence is seen as obstructing or undermining some community dynamic or other, while those suggesting a permissive licence are somehow doing so "in the spirit of Python" (to the point where the inappropriate PSF licence for Python is used for independent projects). > > Well, if you want the users to enjoy those "four freedoms" then you > > should use a copyleft licence. If you choose a permissive licence then > > it more or less means that you don't care about (or have no particular > > position on the matter of) the users being able to enjoy those > > privileges. I believe you coined the term "uncaring", but I think Mr > > Finney's statement stands up to scrutiny. > > I personally don't think that RMS's "four freedoms" are the last word > on the best way for society to develop software, no. But using > "Unless you place such a low value the freedom of your users" is truly > an inflammatory statement, because it was given in a context where the > GPL had not yet been carefully parsed and discussed, and did not make > it clear that the "freedoms" being discussed are a particular set of > "freedoms" and not, for example, those freedoms enshrined in the Bill > of Rights. (And as Steven has carefully pointed out, not all freedoms > are necessarily Good Things.) I tend not to use the terms "freedom" or "right" except when mentioning things like the "four freedoms": the word "privilege" is adequate in communicating what actually is conferred when combining copyright and software licences. Nevertheless, the "four freedoms" and "freedom of your users" are still useful notions: if a proprietary variant of Python became widespread and dominant, although various parts of the software might be freely available in their original forms, the ability to reconstruct or change the software would be impaired and provide fewer opportunities for user involvement than the primary implementations of Python available today. And should such proprietary software become mandated by government agencies or become a de-facto standard, that really does have an effect on the freedom of users. Paul
From: Patrick Maupin on 9 May 2010 18:02 On May 9, 4:21 pm, Paul Boddie <p...(a)boddie.org.uk> wrote: (Lots of good and balanced commentary snipped...) > I didn't say that you personally argued that way, but people do argue > that way. In fact, it's understandable that this is how some people > attempt to understand the GPL - the software maintains a particular > state of openness - but they miss the final step in the reasoning > which leads them to see that the licence preserves a set of privileges > for recipients as well. Obviously, that's a bit subtle and certainly some people might miss the step that says "the final user can always use *my* software, and by choice of license I could decide to insure that if anybody else fixes bugs in my software or combines it with other software they write, then my end users will always be able to use that as well." But others might notice that and still decide that it isn't the right license for their software. > Thus, remarks about Cisco and Linksys - that they were somehow "caught > out" - are disingenuous: if you're in the business of distributing > software, particularly if that software itself has a restrictive > licence, you cannot claim ignorance about licensing or that you just > "found some good code". Any remarks I made about this were only in the debate about "moral hazard" by which I meant something different than Steven did. I'm not at all attempting to condone what Cisco did, just pointing out that if I license code under the MIT license and Cisco uses it, I certainly have no reason to complain about it, and wouldn't dream of doing so in any case, but if I license code under the GPL, then yes, my intentions are clear, and Cisco is complicit in either deliberately turning a blind eye, or inadvertently not watching employees closely enough. I thought I made it clear that I believe that at some level, some human being had to do something that was wrong in order to get Cisco into that position. My sole argument was that in general, I don't want to be the one to put Cisco in that position, and conversely, I don't want anybody else putting me in that position (I want to do the right thing) so I usually don't license stuff under the GPL, or incorporate stuff licensed under the GPL into anything I do. > Why is it pathetic that someone gets to choose the terms under which > their work is made available? By default, if I release something > without any licence, the recipient has very few privileges with > respect to that work: it's literally a case of "all rights reserved" > for the creator. And if it's such a trivial library then why not > reimplement the solution yourself? You just answered your own question. It's pathetic to try to change people's behavior by offering them something worthless if they change their license to match yours. (I'm not at all saying that all GPL code is worthless, but I have seen things like under 30 line snippets that weren't even very well written that were "licensed" under the GPL.) > I dislike the way that when someone releases something under the GPL, > it is claimed that they are coercing or attempting to "leverage" > something. They have merely shared something on their terms. If you > don't like the terms, don't use their software. Well, I don't always make that claim, but I do make it when I see a little "recipe" under the GPL. Often these recipes require so much customization for any particular task that they are really just pedagogical, and some of them aren't even very good. Color me a cynic, but when I see something so short and generic that any sort of judicial test would declare that there was no copyrightable "there" there, that has an arguably politicized license slapped on it, I smell an agenda. > But it is not universally true that GPL-licensed software cannot be > linked to proprietary software: there are a number of caveats in the > GPL covering cases where existing proprietary systems are in use. > Otherwise, you'd never have GPL-licensed software running on > proprietary systems at all. Agreed, but most of those are at very clearly delineated boundaries, like the OS. > > But the tone of your last statement and some of your statements below > > make it abundantly clear that you've made up your mind about my morals > > and aren't at all interested in my reasoning. > > Not at all. Recently, I've had the misfortune to hear lots of > arguments about how the GPL supposedly restricts stuff like > "collaboration" and "growth" despite copious evidence to the contrary, > usually from people who seem to be making a career of shouting down > the GPL or the FSF at every available occasion. Now I'm not saying > that you have the same apparent motivations as these people, but I > maintain that when someone claims that people are "forced" to share > their work when they voluntarily make use of someone else's work, or > that they are at the peril of some "moral hazard", it does have a lot > to say about their perspective. (Not least because people are only > obliged to make their work available under a GPL-compatible licence so > that people who are using the combined work may redistribute it under > the GPL. You yourself have mentioned elsewhere in this discussion one > well-known software project that is not GPL-licensed but was > effectively distributed under the GPL to most of its users for a > considerable period of time.) Sorry, guess I misunderstood where you are coming from. > It is hardly a rare occurrence now that I come across someone who has > written in some corner of the Internet, "It's a shame project XYZ is > GPL-licensed because I can't use it for commercial software > development. Can the project maintainers not choose another licence?" > Sometimes, someone who is seeking licensing advice might not want to > be unpopular and might choose a permissive licence because people > reassure them that their project will only be widely used if the > licence lets people use it "commercially" (or, in other words, in > proprietary software). My impression is that many in the core > community around Python seem to emphasise such popularity over all > other concerns. Yes, but I see the same sort of popularity effects tilt towards the GPL sometimes, too. > What I want to point out, and some have done so much more directly > than I have in other forums and in other discussions, is that some > advice about licensing often stems from a direct motivation amongst > those giving the advice to secure preferential terms for themselves, > and that although such advice may be dressed up as doing the "right" > or "best" thing, those giving the advice stand to gain directly and > even selfishly from having their advice followed. I'm not saying you > have done this, but this is frequently seen in the core Python > community, such that anyone suggesting a copyleft licence is seen as > obstructing or undermining some community dynamic or other, while > those suggesting a permissive licence are somehow doing so "in the > spirit of Python" (to the point where the inappropriate PSF licence > for Python is used for independent projects). Agreed that every project is different, and lots of considerations should be taken into account. > > I tend not to use the terms "freedom" or "right" except when > mentioning things like the "four freedoms": the word "privilege" is > adequate in communicating what actually is conferred when combining > copyright and software licences. Nevertheless, the "four freedoms" and > "freedom of your users" are still useful notions: if a proprietary > variant of Python became widespread and dominant, although various > parts of the software might be freely available in their original > forms, the ability to reconstruct or change the software would be > impaired and provide fewer opportunities for user involvement than the > primary implementations of Python available today. And should such > proprietary software become mandated by government agencies or become > a de-facto standard, that really does have an effect on the freedom of > users. I don't worry too much about that happening. I think that ESR is right that the powerful network effects of free development would overwhelm any attempt to take something like Python private. Regards, Pat
From: Paul Boddie on 9 May 2010 18:05 On 9 Mai, 21:55, Patrick Maupin <pmau...(a)gmail.com> wrote: > On May 9, 12:08 pm, Paul Boddie <p...(a)boddie.org.uk> wrote: > > > Oh sure: the GPL hurts everyone, like all the companies who have made > > quite a lot of money out of effectively making Linux the new > > enterprise successor to Unix, plus all the companies and individuals > > who have taken the sources and rolled their own distributions. > > So, people overstate their cases to make their points. That happens > on both sides. Overstate their cases? The "GPL hurts everyone" is a flat-out falsehood. > > It's not worth my time picking through your "holy war" rhetoric when > > you're throwing "facts" like these around. As is almost always the > > case, the people who see the merit in copyleft-style licensing have > > clearly given the idea a lot more thought than those who immediately > > start throwing mud at Richard Stallman because people won't let them > > use some software as if it originated in a (universally acknowledged) > > public domain environment. > > No, you appear to have a kneejerk reaction much worse than Carl's. > You have assumed you fully understand the motives of people who point > out issues with the GPL, and that those motives are uniformly bad, and > this colors your writing and thinking quite heavily, even to the point > where you naturally assumed I was defending all of Apple's egregious > behavior. I skimmed your post in that particular case and apologised for doing so. How have I not understood the motives of people who do not like the GPL? The GPL sets out a number of conditions on the use of a particular work; these conditions are not liked by some people typically because it means that they cannot use that work as part of a proprietary product or solution, just as the authors of the licence intended; various people would prefer that authors license their works permissively, precisely because this lets them use such works in proprietary software; some of the rhetoric employed to persuade people to permissively license their work involves phrases like "more freedom" (which are subjective at best, although never acknowledged as such) or the more absurd "holy war", evidently. I once attended a talk by someone from the FSF Europe, a few years ago now, where the inevitable assertion that the BSD licences were "more free" was made by an audience member. In my experience, such people are very reluctant to acknowledge the different philosophical dimensions of "freedom", whereas people who apply copyleft licences to their works have typically had to confront such issues even before being asked over and over again to relicense them. > As far as my throwing mud at Stallman, although I release some open > source stuff on my own, I make a living writing software that belongs > to other people, and Stallman has said that that's unethical and I > shouldn't be able to make money in this fashion. Sorry, but he's not > on my side. A lot of people seem to take issue with the GPL because they don't like Stallman, but that only leads to their judgement being clouded as a consequence. When Stallman's warnings become fulfilled, as has been the case with things like BitKeeper, this only serves to infuriate people further, often because they know they could have ignored the messenger but should not have ignored the message. Most people writing software are doing so for other people, and many people are doing so as part of a proprietary process. Thus, the only way to interpret what Stallman has to say (should you not wish to reject it completely) is to consider it as some kind of absolute guidance, not some kind of personal judgement. > > P.S. And the GPL isn't meant to further the cause of open source: it's > > meant to further the Free Software cause, which is not at all the same > > thing. Before you ridicule other people's positions, at least get your > > terminology right. > > And, again, that's "free" according to a somewhat contentious > definition made by someone who is attempting to frame the debate by co- > opting all the "mother and apple pie" words, who is blindly followed > by others who think they are the only ones who are capable of thoughts > which are both rational and pure. I'm not saying everybody who uses > the GPL is in this category, but some of your words here indicate that > you, in fact, might be. No, I am saying that the Free Software movement is a well-defined thing - that's why the name uses capital letters, but it could be called the "Planet of the Zebras movement" for all the name should reasonably distract from what the movement is actually about - and that it has a very different agenda from the "open source" movement which advocates very similar licensing on the basis of things like higher software quality and other "pragmatic" consequences of using such licensing. As for the terminology, I've already noted that I prefer the term "privileges" to "rights" or "freedoms" because it communicates that something is gained. Again, some people assume that the natural state of a work is (or should be) a free-for-all and that the GPL revokes privileges: this is a misrepresentation of how copyright and licensing functions. Paul
From: Paul Boddie on 9 May 2010 19:39 On 10 Mai, 00:02, Patrick Maupin <pmau...(a)gmail.com> wrote: > > You just answered your own question. It's pathetic to try to change > people's behavior by offering them something worthless if they change > their license to match yours. (I'm not at all saying that all GPL > code is worthless, but I have seen things like under 30 line snippets > that weren't even very well written that were "licensed" under the > GPL.) If this is code that you would consider using in an existing project, but if they aren't pitching it directly at you, why would you believe that they are trying to change your behaviour? It is you who gets to decide whether you use the code or not. If the licence isn't acceptable to you, what prevents you from asking for a special licence, especially if you are going to incorporate the code in a product which is sold? In the more general case of people just releasing small programs and libraries, all such people are doing is saying, "Here is something I have done, and here are the terms through which this is shared." If anything, they are reaching out to see if anyone will work together with them on making something better, where everyone agrees to a common framework upon which that work will be done. I'm sure people didn't think much of Linus Torvalds' work in the beginning, either. Paul
From: Patrick Maupin on 9 May 2010 21:09
On May 9, 6:39 pm, Paul Boddie <p...(a)boddie.org.uk> wrote: > On 10 Mai, 00:02, Patrick Maupin <pmau...(a)gmail.com> wrote: > If this is code that you would consider using in an existing project, Well, in a few cases I'm talking about, I wouldn't consider using the code -- I just stumbled across it when researching some terms, and discounted it immediately. Honestly, I'm talking about code that is so small and generic that it doesn't deserve any copyright protection (and wouldn't get any if it came to that in court). > but if they aren't pitching it directly at you, why would you believe > that they are trying to change your behaviour? Because I've seen people specifically state that their purpose in GPLing small libraries is to encourage other people to change their behavior. I take those statements at face value. Certainly RMS carefully lays out that the LGPL should be used sparingly in his "Why you shouldn't use the Lesser GPL for your next library" post. (Hint: he's not suggesting a permissive license instead.) > It is you who gets to > decide whether you use the code or not. If the licence isn't > acceptable to you, what prevents you from asking for a special > licence, especially if you are going to incorporate the code in a > product which is sold? Well, the only time I can remember even hinting around for any kind of different license was when I was found an svglib bug for use with rst2pdf. svglib is a "real" library (unlike the code snippets I was discussing) licensed under the GPL. I would be quite happy to consider learning it and contributing patches to it, but I didn't want to maintain a fork myself, and the maintainer doesn't have a public repository and was quite busy with other stuff, and when I asked him if he would accept patches, it was ten days before he got back to me. rst2pdf was licensed under the MIT license before I started contributing to it, and there is no way I was going to even consider adding patches for a GPLed package (which would certainly have to be GPLed) into the rst2pdf repository. (Say what you will about how sometimes differently licensed code can be combined, but RMS has to share quite a bit of the blame/credit for the whole combining licenses FUD.) So I took a completely different path and right now the best way to use .svg files with rst2pdf is to use inkscape to convert them to PDFs, and use some code I wrote that allows you to use preexisting PDFs as images. Despite being a real library, svglib is quite small at ca. 1200 lines, and if the license were compatible with the rst2pdf codebase license (e.g. even if it were LGPL), I would have just stuffed the file into the rst2pdf codebase and started hacking on it. So there's a nice little piece of GPLed code that isn't getting as much attention as it would have if it were LGPLed or licensed permissively, or even if the author had just dumped it onto sourceforge or googlecode under the GPL but given me commit rights. As it is, I don't think it's been maintained in 8 years. This is exactly the same situation that Carl was describing, only with two different open source packages rather than with a proprietary package and a GPL package. The whole reason people use words like "force" and "viral" with the GPL is that this issue would not have come up if svglib were MIT and rst2pdf were GPL. (Note that the LGPL forces you to give back changes, but not in a way that makes it incompatible with software under other licenses. That's why you see very few complaints about the LGPL.) > In the more general case of people just releasing small programs and > libraries, all such people are doing is saying, "Here is something I > have done, and here are the terms through which this is shared." Sure, and all I'm explaining is why I reject the terms in some cases. I particularly reject the terms when a license that was originally designed for a whole program is used for a small library, with the express intent of getting more people to use said license. > If anything, they are reaching out to see if anyone will work together > with them on making something better, where everyone agrees to a > common framework upon which that work will be done. That's absolutely not always the case. Often, it's more "here's something I've done; take it or leave it but if you take it, it's on these terms." > I'm sure people > didn't think much of Linus Torvalds' work in the beginning, either. I think the GPL was a great license for the development model for Linux. I have no issues with how that worked. Part of that is that Linus was always active in the development. I think that, particularly when the concept of free software was relatively new, the license might have been an effective focal point for rallying contributions. But I have definitely seen cases where people are offering something that is not of nearly as much value as they seem to think it is, where one of the goals is obviously to try to spread the GPL. Regards, Pat |