{"thread":{"id":"16345","subject":"Yes, SymPy should be LGPL! (please help)","startedAt":"2008-11-16T00:07:53Z","lastAt":"2008-11-16T00:07:53Z","messageCount":1,"participants":["Kirill Smelkov"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"95913","messageId":"20081116000753.GA9636@roro3","threadId":"16345","inReplyTo":"20081115114859.GB23252@phare.normalesup.org","subject":"Yes, SymPy should be LGPL! (please help)","fromName":"Kirill Smelkov","fromEmail":"kirr@landau.phys.spbu.ru","sentAt":"2008-11-16T00:07:53Z","receivedAt":"2008-11-16T00:07:53Z","isPatch":false,"sender":{"key":"kirr@navytux.spb.ru","avatar":"https://gravatar.com/avatar/cf3445fdad1849941e17ab25bf1ee7c5ea1be2deb444be25ff5f36b0e50a985f?d=mp&s=160"},"body":"\n[First of all, sorry for off-topic to those lists not related to SymPy]\n\n     This email is posted here in hope to reach people who share my\n     point of view, and to ask to please help to\n\n     _____________________________________________________\n    /\\                                                    \\\n    \\_| Advocate switching SymPy license from BSD to LGPL |\n      |                                                   |\n      |   ________________________________________________|_\n       \\_/__________________________________________________/\n\n\nStart of this thread is here:\nhttp://groups.google.com/group/sympy/browse_thread/thread/b5addd939ee0c803\n\n\nIntroduction\n============\n\nCurrently SymPy is distributed under new BSD license.\n\nBut this was not always so -- originally the project was licensed under\nGPL (!), but in the beginning of 2007, after Pearu Peterson has proposed\nto switch to LGPL or BSD [1], developers decided to take permissive\napproach -- BSD.\n\nI've joined SymPy project later - in 2007 summer, and now I'm\nproposing we change SymPy license to LGPL (Lesser GPL) which I\nsee as a golden median between beeing too permissive (BSD) and too\nstrict (GPL).     |\n                  |\n                  v\n        BSD ---- LGPL ---- GPL\n\n      +------------------------->\n                       strictness\n\n\nRead below why\n\n[1] http://code.google.com/p/sympy/issues/detail?id=138\n\n\nOn Wed, 12 Nov 2008 13:24:39 -0800, Robert Kern wrote:\n> FWIW, my vote is for BSD. I, personally, don't work on (weak or\n> strong) copyleft projects. Copyleft licenses are intrinsically\n> attached to an agenda which I don't share so I don't use them. I'm not\n> a fan of agendas.\n> \n> I can see the point for some pieces of software where there is a known\n> threat (e.g. the LGPLed work on GMP gets enhanced slightly by a\n> proprietary product and resold with lots of bragging about how they're\n> better than the open source GMP), but is there an actual threat to\n> sympy? Hypothetical threats don't impress me much.\n\nReally?\n\nEven if it is not evident to the casual observer, in the begginning of a\nchess play, there is already a threat to your figures, and your king and\nqueen, so chessplayers usually try to analyze the situation early and\nlook into the future for at least several turns. And the player, who can\nanalyze more turns than the other, wins.\n\n\nMy understanding is that SymPy is close to enter \"serious CAS\" market,\nand as Ondrej says below, SymPy has a lot of potential. To me what is\nneeded to be done for this is to go into compiled mode for core, and to\nrefine core design and it's data structures. Plus constantly trying to\napply SymPy to real-world problems, so that we could be sure we have\ngood engine connected to good user interface.\n\nAs during the last year I've touched both internal and high-level stuff\nquite a bit, I think I'm keeping my hand at pulse here, and my\nunderstanding is that together with all the polishing, assuming the\nprevious temp of work, this transition should take 6 - 12 months.\n\nAnd after this transition is done, wouldn't it be like the case with\nGMP, in which you say you see the points for LGPL?\n\nOnly then it would be late (or much more difficult than today) to protect\nSymPy from abuse -- That's why I care *now*.\n\n\nOn Thu, 13 Nov 2008 01:54:23 -0800, Andy Ray Terrel wrote:\n> This is something of a tough discussion for me, mostly because I\n> disagree and agree with a large number of points made here.  I\n> ultimately reject that the 4 freedoms of Richard M. Stallman (RMS) are\n> some how fundamental to a computer user.  I prefer the arguments given\n> by Lawrence Lessig, that a free culture is a better more productive\n> culture.  RMS does make a very nice point that a way to preserve this\n> free culture is to practice the 4 freedoms and require those who use\n> our software to comply.  Lessig's Creative Commons licenses allow for\n> this structure as well as others.  Nevertheless I also hold firmly\n> that it is rather the community of developers and not the legal\n> structure it stands on that make open source such a great experience.\n> I and many friends have developed BSD, GNU, and proprietary licensed\n> software with similar goals to promote technology and human progress,\n> much less motivated by secret agendas or monetary reward.  A lesson I\n> have learned, projects that are *only* about money are never the\n> innovators of technology.\n> \n> The debate I do agree with RMS on is in the view of labor rights of\n> the computer programmer.  Programmers are not the first community of\n> artisans where it is easy to manipulate their works for personal gain\n> rather than that of the community.  By following a copyleft license\n> the computer programmer is investing in her community and protecting\n> that investment from those who would falsely claim the work as their\n> own.  Agreeing with Linus Torvalds, this does not mean to me one\n> cannot work with proprietary projects but rather the investment in the\n> community creates better software.  The BSD license protects the\n> developer, the GNU set of licenses protect the community.  I find a\n> nice balance in the LGPL license because it does allow proprietary\n> development and marketing but still protects the community.\n\nYes, in essence, this is what I'm advocating for:\n\n    a nicely balanced LGPL license which allows proprietary development\n    and marketing but still protects the community\n\n> You are going to lose users for many reasons. Losing them because do\n> not wish to conform to policies in place to protect the creative\n> investment of the community is not a bad reason.\n\nYes, users and developers come and go.\n\nA lot of people listed in our README already went away for various\nreasons. Some did not fit into team-development and respect-each-other\nand dont-break-existing stuff rule, some got afraid of patch review (I\nsuspect I was the main source of disaster), some got busy with other\nstuff, etc...\n\nIt is not possible to be good for everyone -- that's why I don't try to\nsatisfy all the people, but instead try to be good for me myself, and do\nwhatever I think is better for me and for SymPy code, and if the result\ncould be somehow useful for others - I don't mind at all.\n\nI hope that the last year shows that this approach worked quite well.\n\n> I think sympy is a great community not because of its license but\n> because of the people and I will continue to invest in such a\n> community no matter what the license.\n\nLikewise did I.\n\nI've put licensing stuff on shelve in harder times, just to concentrate\non technical bits and defend SymPy. But after that investment (it was in\nspare time + several day offs, compare with e.g. 6 months full-time\ngrant sponsored development of sympycore [1]), it became much more evident\nthat developers effort have to be protected.\n\nYes, to me developers effort have to be respected and protected. And if\nthis could be done in a compromise way, it should be done.\n\nPlease read below, where I'm advocating LGPL in more details.\n\n[1] http://code.google.com/p/sympy/wiki/SymPyCore\n\n\nOn Thu, 13 Nov 2008 09:27:29 -0800, Vinzent Steinberg wrote:\n> First of all I want to thank you, Kirill, it was great to enjoy your\n> support. You are certainly one of the most constructive persons I\n> know.\n\nThank you Vinzent for you kind words. This proposition gives me one more\nreason to think that this time, I'm too proposing something\nconstructive.\n\n> I think the problem is quite fundamental. Many developers are\n> investing quite a lot in an open source project, and they don't want\n> others to modify it without profiting from their modification. This\n> point I perfectly valid.\n> \n> Given that sympy would be picked up by an company and they would turn\n> it into a closed source mathematica2, some sympy probably developers\n> would say: \"Great, this draws much attention to our open source\n> project, and we will profit from the research they will do, because we\n> can see how they solved problems, even if they don't share their code\n> with us. And after all, it's better they invest in this area without\n> sharing code than not investing anything (due to restrictive\n> licenses).\" For this kind of developers, the BSD license is better,\n> because they think it's better for the project and the code in their\n> opinion.\n\nI'll tell you one story about me and physics:\n\nI was working on some physics research and developed my own routines\npack for this. At seminar, after I've talked about current progress, one\nof my collegues asked me:\n\n    \"Kirill, you have done nice job with developing you routines pack,\n     may I borrow them for my research too?\"\n\nSure, my answer was\n\n    \"Yes, please. Here you are.\"\n\nOn the next seminar that collegue was talking about his progress, and he\nmentioned my routines, and that he enhanced them a bit.\n\nThis time, I've asked him:\n\n    \"Nice job, could you please give your versions of my routinse to me?\n     I'd like to use them too\"\n\nand, the answer was\n\n    \"No, I can't. I need the result to be published first, so that I\n     have acknowledgment for my research. After I'll publish the\n     results, I'll give you your original routines enhanced by me\"\n\nI thought.\n\n    \"Ok. I was not going to compete with this guy and steal his credits\n     in the first place anyway, but it seems he wants to be extra sure\n     in this, so even if it is a bit pity, that I can't use the\n     enhancements yet, so be it.\"\n\nTime has passed, and on the next seminar, my collegue was talking about\nhis new progress, and mentioned his recently published paper with great\nresults obtained with the help of my routines. It turned out that this\ntime, they were again enhanced slightly more.\n\nI've asked him:\n\n    \"Now you have your paper published, could you please give me my\n     enhanced routines, so that I can use them for my research too?\"\n\nI was sure, he would say \"yes, please\", but I've made a mistake:\n\n    \"No, if I give you your enhanced routines, this would make us equal,\n     and I could not be sure to win this tasty new grant we'll be\n     competing for on the next week.\"\n\nI thought huh!? How is it possible that I've shared my research in the\nfirst place, and now he says he is not going to do the same, even when\nhis work stood on my basis?\n\nNeedles to say, that the next time, when I talked about my progress and\nnew software which I had to develop along the way, I refused to give it\nto my collegue, because I though that he plays unfair:\n\n    \"Last time you took my routines and enjoyed using them, but when I\n     asked you to share your enhancements, you refused for various\n     reasons.\n\n     So why should I give my work to you this time?\"\n\nThis conflicting situation became known to our management, and they\ntried to analyze it and asked me:\n\n    \"Kirill, at our department we do different kind of research, and we\n     have lots of groups. How we work was always cooperative in spirit -\n     we always used to share the results internally and give advices to\n     each other.\n\n     Yes, sometimes there is concurency between groups, but this was\n     never an obstacle for cooperative research, because everyone is\n     aware and confirming to good science ethics - not to publish\n     overlapping results before the original author.\n\n     Why are you refusing to share your work with collegue?\"\n\nI told them that we used to share codes, but somehow contributing flow\nbecame unidirectional, and that I think this is not fair to only take\nand not to give back:\n\n    \"I'm ok with my collegues developing their own private codes, even\n     if they are using mine, but I'd like them to share back their\n     changes to _mine_ code. It's _their_ behaviour which is not\n     cooperative.\"\n\n\nThis story is imaginary, but I think it fairly well illustrates\nqualitively the situation I have in mind. You can think of two\nphysicians, two mathematitians, or sympy development community v.s. some\ncompany which ships locked down sympy to its customers.\n\nAnd it's this \"we only take and locally modify\" which I propose to\nlegally prevent.\n\n\n---\n\nAs to\n\n   \"Great, this draws much attention to our open source\n    project, and we will profit from the research they will do, because we\n    can see how they solved problems, even if they don't share their code\n    with us. And after all, it's better they invest in this area without\n    sharing code than not investing anything (due to restrictive\n    licenses).\"    \n\nI disagree that in the long run, the original project will profit.\n\nPractice shows that instead, usually it's closed neighbour draw\ndevelopers (key developers are usually just hired) and users, but after\nthis happens, the main project stays with even less resources, and this\nstrikes back at both open and closed projects. End of story.\n\nPlease read below about how this happens.\n\n\n> But others would mention: \"I want my project to be open, closed source\n> is completely useless for me. I spent my work to let others share, not\n> that someone grabs it to make money with it without giving anything\n> back to the community. Freedom is one of the most important things\n> about our project, and this company is destroying freedom. Because\n> they have more resources they will have a more polished product and\n> drawing off our users and contributors. It's ok if they use it to earn\n> money, but they have to give something back\"\n> \n> These opinions are quite polarized and are probably not well\n> formulated by me. Mainly I want to say that both opinions are equally\n> right. It's up to everyone to decide which license and use of his very\n> work pleases him most and to respect the decisions and motivations of\n> developers preferring other licenses. You can't make both groups\n> happy.\n\nYou *can*. For example glibc being mostly LGPL is used by every program\nand every library on Linux system, including e.g. Python which is\nBSD-like licensed, and every proprietary program which runs on GNU/Linux.\n\nMy point here is that LGPL (Lesser GPL) is a good compromise between\nbeeing too strict (GPL) and too permissive (BSD).\n\nActually my license would be the \"Fair use of SymPy\"\n\n\"\"\"\n    Fair use of SymPy\n    -----------------\n\n    You can use SymPy as you want, and we encourage you to build various\n    products on top of SymPy including commercial and closed-source ones.\n\n    Your own code that uses SymPy as a library can perfectly stay\n    proprietary.\n\n    But if you change SymPy itself -- you have to share your modifications\n    to SymPy, your other code still stays yours.\n\n    To legally state what we think is our \"Fair use of SymPy\" we adopt LGPL\n    license.\n\"\"\"\n\nAs last line says, its simply LGPL who is the closest-in-spirit\ntranslation of this terms which is composed by legal-aware people and is\ntested.\n\n> An example for such an scenario is Wine. They switched from MIT to\n> LGPL due to a company not contributing changes back to the core\n> project.\n\nYes, Wine is a good example.\n\nAnd before switching, they were too wanting \"Wine to be used as wide as\npossible\" - that's why they were licensed by MIT (similar to BSD).\n\nBut when they started to be abused by TransGaming (who developed closed\nWineX and did not contributed back their changes), they decided to\nswitch to LGPL, because LGPL'ed Wine is still perfectly allowed to be\nused by essentially any software (closed and open), but changes to Wine\nitself have to be always made shared with whom you distribute Wine.\n\nhttp://wiki.winehq.org/FAQ#head-72731b215bbd1ce36e3b84ac7ce114925ce16460\nhttp://wiki.winehq.org/WineHistory\n\nSo they changed from BSD to LGPL and this is what happened:\n\n\"\"\"\nThe immediate effect was the creation of the ReWind project to further\nthe X11-licensed codebase. Many key developers agreed to allow their\nadditions to be used by both the X11 and LGPL Wine code. Most have\ndecided to focus their efforts on synchronizing with the LGPL'ed Wine\nand the vast majority of development and new features appear there\nfirst. Picking a favorite software license is left as an exercise for\nthe reader.\n\nShortly after changing the license development began to pick up at a\ngreater pace. More patches began to appear, Alexandre made more CVS\ncommits, and more applications were reported to work. By the end of\n2003 DLL separation achieved a milestone with the split of the\nkernel32/ntdll combination. Most of the architectural changes to\nsupport a beta release were in place. Another developer's conference\nwas held in January, 2004 in St. Paul Minnesota sponsored by\nCodeWeavers. Once again, a roadmap was laid out for tasks that needed\nto be completed.\n\"\"\"\n\nSo it seems LGPL also boosted Wine development.\n\n\n> I would count myself to the first group (I wouldn't mind if a company\n> would do such things to sympy), but I comprehend very well others like\n> Kirill who do not want this.\n\nYes, I strongly mind when someone only takes, modifies, and gives\nnothing back. As said above, I'm ok if anyone takes SymPy and uses it in\na closed-source proprieary product, but if he changes SymPy itself, I\nstrongly mind for that SymPy-changes to be closed.\n\n\n> > \"make private modifications to sympy and keep them secret.\"\n> \n> If I understand the GPL correctly, it won't prevent this. It does not\n> force you to give modifications back. It only forces you to distribute\n> your program with sources. This means they can keep modifications\n> private as long as they don't sell the software. And this practically\n> means that modifications won't be secret of course. :)\n>\n> I don't like that GPL is so viral. It's often incompatible to other\n> open source licenses. But it's a great choice if you want to to avoid\n> these \"private and secret modifications\".\n\nPlease, why are you talking about GPL here?\n\nI'm proposing LGPL which is a different, \"non viral\" license [1].\n\nWhen you mix GPL here, you create confusion, and I think that confusion\nis the first step towards creating FUD (Fear, Uncertainty and Doubt)\n[2].\n\n\nAs to you statement - yes, I don't care about someone keeping private\nmodifications without redistributing them -- all I care about is\nredistribution, and this is covered by LGPL.\n\n\n[1] http://en.wikipedia.org/wiki/GNU_Lesser_General_Public_License\n[2] http://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt\n\n\nOn Sat, Nov 15, 2008 at 12:48:59PM +0100, Gael Varoquaux wrote:\n> On Wed, Nov 12, 2008 at 09:47:18PM +0100, Ondrej Certik wrote:\n> > it's a very nice and honest email. Kirr has done a stellar job over\n> > the past year and half and we definitely would not be as advanced with\n> > sympy without him. I'll reply to his arguments in my next email.\n> \n> I'll use my mail to reply mainly to Kirr (I hope he is still listening)\n> because I hugely respect and value a man who has invested a large amount\n> of quality time in a community project.\n> \n> Kirr, you have been a fantastic actor of the project. You are now saying\n> that you are tired. This is actually a fairly common problem. We all\n> fight with lack of time, to many various interests. Software development\n> comes with its load of difficulties and disappointments. Too much\n> sacrifices ruin private life.\n> \n> I am off course very sad if you quit this project that I follow (without\n> contributing, sorry) and in which I believe. However, what I really want\n> is to thank you for all your work. I hope you had as much fun as\n> possible, and I hope that your will keep good memories of your time with\n> sympy, rather than the reasons that made you quit. And besides, no\n> decisions is irreversible. I believe that you will have friends in the\n> sympy team whether you are contributing or not. Life and friendship are\n> more important than software. \n\nThanks Gaël for your words too.\n\nBeing friends is one thing, protecting the results of hard work from\nabuse is another one. That's why I'm here.\n\n> > But I'd like to use this occasion to make a survey among our\n> > developers and users ---- do you think we should stay with BSD, or\n> > switch to LGPL?  I am interested in all opinions from anyone. Whenever\n> > there is some occasion, I am discussing this with people around sympy\n> > and so far my observation is that most people either prefer BSD, or\n> > don't mind either, but would like to hear good arguments why LGPL\n> > should be better. There are also some who clearly prefer LGPL, for\n> > example Kirill or some people from the Sage community, but I think\n> > given all the circumstances (i.e. sympy aspiring to become the\n> > symbolic manipulation library for Python, and together with numpy,\n> > scipy, matplotlib, mayavi and others form a BSD like environment for\n> > scientific computing, that people can build on and given the community\n> > where sympy belongs), the right license for sympy is BSD.\n> \n> We have discussed this a few times, including over a few beers in Prague,\n> you know my point of view. I won't dwell on it long.\n> \n> Briefly. I personally realized that I don't care much about the license\n> of the software that I write. I care more that it is used by as many\n> people as much as possible. The thing that worries me most with software\n> is the amount of wheel reinvention, and lousy, half-complete,\n> implementations of a solution to a given problem. This is why I believe\n> free software has a future. This is indeed the Lessig point of view. The\n> reason I feel a BSD license help is that many companies (especially large\n> ones) are not afraid of hiring scores of developers to reinvent the\n> wheel, just because they want to possess it and control it. The driver\n> for free software adoption in companies is not cost, but control. \n\nYes, usually companies wants control, and usually that control hurts the\nproject. That's why I think that SymPy should be independent from any\ncompany at any give time.\n\nPlease read below about it.\n\n> Of course the license debate changes whether you consider your software\n> as a component, that needs to be integrate in a larger infrastructure, or\n> as a finished good.\n\n> Linux, (GPL, fantastic example of free software success) is clearly a\n> finished good.\n\nYou forget that it was not \"finished good\" from the day0, but it was GPL\nfrom almost the begginning. Also, please note, that in Linux kernel case\nGPL works essentially like LGPL -- please read about it below.\n\nSo it *evoloved* to \"finished good\" with the license it carries today. And\neven before Linux was _started_, FreeBSD & friends were already there,\nright?\n\nHow it turned out that today Linux outperforms (my subjective point of\nview) FreeBSD with respect to almost all areas? For example FreeBSD had\nmuch better networking stack than Linux some years ago, but today the\nsituation is different.\n\nDoesn't it points that somehow the license is too mixed into success?\n\nTo me, the answer is \"yes\" - read about below why.\n\n> Wine (now LGPL) may also be a finished good.\n\nThe same points as with Linux kernel - it too was not \"finished good\" from\nthe start, plus as said above Wine switched to LGPL to protect itself\nfrom abuse back in 2002.\n\n\nAnd even today, both Linux and WINE are _not_ finished goods.\n\nThey are developing at a very high pace - especially Linux. For example\ntransition from 2.6.26 to 2.6.27 which lasted for ~ 3 months brought a\nlots of new features [1] and had ~ 10000 (ten thouthands) commits (!)\n[2]. Compare this with SymPy, which has ~ 2500 commits for _all_ its\nhistory! Roughly saying every new kernel version is like ~ 3-4 man-work\nof the *whole* sympy development and this is even more, if you'll bear\nin mind, that the kernel patch quality rules are much higher than in\nsympy.\n\nI'd not call it as \"finished good.\"\n\n[1] http://kernelnewbies.org/Linux_2_6_27\n[2] http://lwn.net/Articles/302061/\n\n> Scientific computing libraries on the other hand, are not\n> finished goods. A company, or a research lab, may want to fork them to\n> keep a competitive advantage, and maybe not even redistribute the\n> resulting binaries. And that's fair enough by me, because I think the\n> world is a better place if these people use my software as a start,\n> rather than building on crappy bases.\n\nAs pointed above \"finished good\" or not is not an argument for me.\n\nBesides, don't this people constantly reinvent \"lousy wheels\"? And why\nthe world is a \"better place\" with those \"louse wheels\" forked from your\nbase.\n\nWouldn't it be better to cooperate on common playground?\n\n> besides, if they later decide to\n> share, it will be possible if the built on common abstraction. Finally, I\n> hope that as the world evolve, people (mostly managers and lawyers) will\n> understand the gain that there is to share 90% of the code, may it be\n> with 10% kept secret for a competitive advantage.\n\nMy point is that we should not wait untill those managers understand\nsomething (or not), but create a world in which a manager that does not\nunderstand this something, can't effectively manage, so he/she will just\nhave to educate him/herself.\n\nRead about this below.\n\n> For me the golden example of successful free scientific software is VTK.\n> VTK is a 3D plotting library. 3D plotting is a forever-reinvented\n> problem. Each particular application is actually fairly simple, but when\n> you try to do something a bit more general, it gets really hard. So\n> people (companies, research labs) start out with a specific problem they\n> want to solve, and code a quick implementation, then as they evolve,\n> their small implementation grows to be a huge one, and most of the time\n> lies on crappy abstractions and data structure. VTK was born as a spin\n> off of GE: the code used in medical visualization was split off to a\n> start up (Kitware) and BSDed. GE still have their own proprietary bunch\n> of codes, for their medical devices, but VTK lives on, and unites the\n> efforts of many companies and labs with diverging problems and goals. In\n> the case of VTK, individual actors have learned that they need to\n> contribute back, because the cost of have their own abstraction and\n> algorithm to maintain and to impedance-match with the rest of the world\n> was very high. They make a careful decision as to what is open-sourced\n> and what isn't.\n> \n> This is my model of development, and my 2 cents.\n\nI can't say about VTK, because I don't know this area, and also I admit\nthere are good projects under BSD license but wouldn't the same work if\nVTK would be LGPL'ed?\n\nLet's forget for a moment that some companies are afraid of *GPL, and\ntry to see -- what would be an obstacle to keep VTK, which is open,\nunder LGPL, and to keep private additional development private?\n\nFor me it's better to have clear rules, what _have_ to be always common\nand belong to everyone. When such rules does _not_ force other\nnot-covered components (e.g. an aplication which uses sympy) into which\nlicense to use, I call it non-intrusive.\n\nAnd to me, LGPL is non-intrusive (read about glibc, libgcc, and wine\nv.s. picassa) and it protects developers. That's why I propose we use\nit.\n\n\nOn Wed, 12 Nov 2008 13:59:07 -0800, Ondrej Certik wrote:\n> On Wed, Nov 12, 2008 at 8:44 PM, Kirill Smelkov\n> <kirr@landau.phys.spbu.ru> wrote:\n> > Compared to bad science, or locked down or proprietary software, it is a\n> > very good thing!\n> >\n> > I'd like to explain better, but now I'm too sleepy, and I have 10\n> > minutes left, and tommorow I have to go to work and concentrate there\n> > ...\n> >\n> > <so from now on it goes unstructured>\n> >\n> >\n> > What is important is that the software should be free as in free speech\n> > - one should always be able to:\n> >\n> > - run it\n> > - study it\n> > - modify it\n> > - share it\n> \n> Yes. You should also be able to compile it (=run it) using only free\n> software. E.g. that's what Debian main distribution preserves.\n>\n> > There are mechanisms to legally make contracts with software users,\n> > which protects this freedom, mainly strong copyleft (GPL) and weak\n> > copyleft (LGPL).\n> \n> It is important to note, that not even GPL protects all kinds of\n> \"abuse\". In my opinion, the far bigger \"threat\" are the internet based\n> applications, like gmail, that are completely without sources and the\n> companies can even run GPLed code behind it --- but you have no chance\n> to study/run/modify or share the program.\n> \n> But as long as I am not forced to use such a service, I don't mind.\n> And I use gmail, because it's good. But I don't have to, I can\n> administer my own email server and use mutt. As long as I have this\n> second option, I don't mind using non-free software (I also sometimes\n> use google earth and skype and picasa and other things). I just don't\n> want to depend on it.\n\nYes, such kind of thing exists. Let's try to agree on our terms \"in\nspirit\" first -- if we do - we'll work all the technical and wording\ndetails out. Is my \"Fair use of SymPy\" ok? If so, let's just adopt\nclosest existing solution on license market which is not-intrusive.\n\nAgree?\n\n\nNB: Also I think that every time you say \"I can\" (administer your own\nmail server, use mail client, ...), and someday you need to turn some of\nthis points into reality, multiply the probablility that the whole thing\nwill happen, say by 0.1 for each 'can'. Then you'll understand, that\nwhen there are too much this \"can(s)\", what you get in essence is that\nyou can't.\n\n\n> > From almost the beggining I was talking to Ondrej that I think it's\n> > better to use LGPL for SymPy, that is anyone could be able to build\n> > application or other libraries whichever license they choose (including\n> > proprietary) on top of SymPy, but SymPy should always stays free as in\n> > free speech. Mainly one should be disallowed to \"make private\n> > modifications to sympy and keep them secret.\"\n> \n> Here we disagree. For me personally, this right is very important,\n> that not only I, but anyone can make any modification and do what he\n> wants with it, including keeping it secret.\n\nWhat's your rationale on this? Why you need the right to keep\nmodifications of sympy secret? Is there any reason this is good for\nsympy?\n\nAs I've stated above I strongly mind about secret modification which are\nredistribted in closed form, and I think this is not fair.\n\nRead below about my understanding of why you need to reserve such rights\nas \"keeping modifications secret.\"\n\n> > LGPL is a good compromise, and it does not force users of a library into\n> > any license, so why can't it be used.\n> \n> Well, it depends on your personal point of view, but many people,\n> including me view it in a way that it forces, because LGPL is not\n> compatible with every license out there. Actually, I think there are\n> even some problems with mixing LGPL 2 and LGPL 3 code.\n\nI'm not a lawyer, but as I've said above, isn't GLibc (coverd by LGPL),\nwhich is used by essentially *every* program, is not a problem? What\nabout libgcc (GPL + linking exception) ?\n\nI insist on that we should settle on \"Fair use of SymPy\", and if we do,\nit would be straightforward to adopt the closest license out there on\nthe market, and work out all the details to eliminate license\nincompatibility out there if any.\n\n\n> > Why on earth, say I as a hobbyst, spending my spare time, would want the\n> > result be used as a starting ground for closed software? BSD allows it\n> > explicitly, and this is not ok with me.\n> \n> I respect your point of view and I think it's perfectly consistent.\n> But I don't share it --- I am also a hobbyist to some extent, well as\n> to sympy I still am, and I just want my code and my work to be used.\n> Because I believe it is good for me and the whole community in the\n> end. I do sympy because I very strongly believe in its potential and\n> that we get to a point (and we are almost there) where I could use it\n> regularly for my research. For example our recent work with Brian\n> shows, that we could actually do atomic physics using only open source\n> software in Python. That's very exciting.\n> \n> And if someone uses it in some commercial product? I believe it will\n> help us, not hurt us. I believe it's about the community which is\n> around sympy. Commercial product can have a community, but a different\n> one --- I am pretty sure we all use sympy exactly because we have the\n> source code and because it's opensource.\n\nCommercial != Closed\n\nYou can build a commercial product on top of open development. An\nexample of this is CodeWeavers (CrossOver -- WINE) and RedHat (RHEL --\nFedora)\n\nI think commercial parties are ok, and essential for success, but the\nwhole thing works well, when there is no control in one hand, and when\nparties (commercial or not) play by the same rules.\n\nAlso, since I propose we use LGPL, it would be perfectly possible to\nbuild proprietary/closed software on top of sympy.\n\nWhat's the problem with this?\n\n> > Actually at my work, With one of my collegues, we've made an\n> > observation, that usually motivated people, who are interested first in\n> > developing interesting things, to achive great result, to care about the\n> > project and it's ecosystem in general, and only then in money, usually\n> > prefer LGPL to BSD, and on contrary side, people who tend to work for\n> > money first, and care less about result, about fitting their development\n> > with others, about achiving good results, pushing harder when needed,\n> > prefer BSD. They usually say that \"BSD is the license that companys\n> > prefer, and for which they pay money\"\n> \n> I think both ways are possible. I believe that Python itself and all\n> the BSD like tools around it show that people can do BSD software and\n> care about the project.\n\nThere are exceptions from the pattern, and examples of both good BSD and\nbad LGPL projects.\n\nEvery project is somewhat unique with additional balancing forces (e.g.\nPostgreSQL - read below), but to mee the tendency is clear.\n\n\n> > Do you see the pattern?\n> >\n> > LGPL, being a fair contract, is good for everyone -->\n> \n> Here we disagree. I respect that for you \"LGPL is good for everyone\".\n> But for me, \"BSD is just as fair for everyone\".\n\nHow is it fair, if let's say I've developed sympy for a year in my spare\ntime at evenings, and someone would just take it, work on it full-time\nand enhance it, and sell it, and keep the modifications secret?\n\nIt seems we have different understanding of fairness.\n\nAnd what about GLibc which is LGPL'ed? Any problems in using this\nlibrary? I bet 90% of people out there even dont _think_ of this issue\nwhen they run their programs, that they depend on Glibc.\n\nAnother news: libgcc (The GCC low-level runtime library) is licensed by\nGPL (!) + linking exception [2]  and is used by every program which is\ncompiled with GCC !. No problem with this???\n\n[1] http://gcc.gnu.org/onlinedocs/gccint/Libgcc.html\n[2]\nhttp://git.infradead.org/gcc.git?a=blob;f=gcc/libgcc2.c;h=5a82f82f7cd2f2e2ff8a19b157ec13529b3d8251;hb=HEAD\n\n> > (e.g. GTK+ is LGPL'ed, and there is a lot of free and proprietary\n> > software which uses it without a problem)\n> >\n> > --> is choosen by people who care, whereas BSD is choosen by people who\n> > want money first.\n> \n> I understand what you mean. But I don't think that people around\n> sympy, or scipy or numpy don't care and want money first. So I\n> disagree with that statement.\n\nActually some people we all know want money first [1], or stop working\non their \"babies\" after their grant is finished [2], or want to be paid\nworking for companies which want to keep their modifications secret [3],\n...\n\nOther's (usually more young) just don't understand the issue and don't\ncare about licensing --\n\n-- BSD is essentially a gift, and you know when you are a student you\ncan afford yourself to give gifts whatever way you like, and also there\nwas a culture in the university to share things \"simply\" -- this is all\ngood. -->\n\n  HISTORICAL REMARK: BSD-like licenses were born in the universities,\n  where the research and development were done on taxpayer's money, so\n  it was logical that the result belongs to all the taxpayers --\n  basically everyone -- commercial and open parties -- all the people.\n\n  That's why BSD-style licenses are so permissive -- actually BSD, MIT\n  and the like are very close to Public Domain.\n\n\n  Today most of the development is not done on taxpayer's money, so\n  likewise, there is no obligation for the result to stay close to\n  Public Domain -- that's why usually software developers who do the\n  actual work in their spare time are those who decide on which terms\n  they distribute the software.\n\n--> But we are not in the university where there are > 95% good people, and\nwhy would I want some bad guy to use the code I spent my time on to use\nin a bad way? I say - 'use it as you like, but if you enhance it, you\ncannot enjoy your advantages alone - you have to share this'. I think\nthis is 100% fair and thus LGPL is 100% ok with me.\n\n\n\n[1] http://code.google.com/p/sympy/issues/detail?id=748#c25\n[2] http://landau.phys.spbu.ru/~kirr/cgi-bin/hg.cgi/sympycore--hg/\n[3] http://groups.google.com/group/sympy/msg/74a5c4bc7506ab68\n\n> > Money is ok, but I even can say that usually who want money first, get\n> > _less_ money than people who do care about the project in the first\n> > place.\n> \n> Yes, I agree. I believe in american dream which says that you can get\n> what you want, but you need to work hard and it may take a long time,\n> but you'll get there eventually.\n>\n> > Also I can say that in our company we use a lot of LGPL'ed software, and\n> > this is legally 100% ok, and that we even sponsor some LGPL'ed\n> > development, so to me there is just no rational reason why LGPL is not\n> > good for SymPy.\n> \n> There are pros and cons to all three BSD and LGPL and BSD. You stated\n> the pros of LGPL. The cons is that LGPL creates a barrier of\n> contributions, from a lot (but not all) people and also usage, because\n> simply LGPL is more restrictive.\n\n1. What is the usage barrier of LGPL? I remind you about GLibc, and\n   libgcc. Please answer in details about this barrier - to me it is\n   mythical.\n\n2. What is the contribution barrier of LGPL? I know there are people who\n   hates GPL (no L) and automatically projects their attitude on LGPL,\n   and also that there are companies who like BSD and cultivated this\n   (read about it below).\n\n   But what rational is in this barrier?\n\nAlso, as you say that LGPL creates barriers, I state that BSD *TOO*\ncreates barriers:\n\n    Look, they say GPL is viral, and some time ago I undersood that BSD\n    is viral *too* -- it forces all to use BSD -- look at scipy -- they are\n    reluctant to take LGPL parts even it it is (maybe) legally ok to ship\n    combined BSD + LGPL library (part1 is BSD, part2 is LGPL)\n    \n    so BSD creates barriers TOO.\n\nThis is also justifed by people like me, Gary from SAGE and others\nrefusing to contribute to a BSD licensing projects.\n\n> Whenever you put more restrictions (I\n> am not saying it is necessarily bad) you lose some users. So while\n> LGPL is maybe good for sympy, BSD is imho better.\n\n1. As shown above LGPL does NOT create barrier for _users_,\n\n2. With BSD you loose (at least) some developers (me, Gary, other SAGE\n   people)\n\nSo why is BSD better, when LGPL being almost the same for users,\nprotects _developers_? Because companies can't \"keep their modifications\nsecret\"?\n\n> > I've failed to win Ondrej's mind, and I think I'll fail to convince most\n> > of you, and it makes me very sad, that so much evident common sence is\n> > not accepted, that I stopped thinking to be like at home with sympy\n> > recently.\n> \n> I am sorry about that. But I think it is not a common sence.\n\nI insist on it being common sense.\n\n\n> > either having a rest, going for a walk, for a talk with a human you\n> > love, or either to develop sympy, is _precious_.\n> \n> I agree with this too. We all die once, so our time is limited and so\n> we should spend it very effectively.\n\nYes, but effectivness depends on what you trying to achieve. If someone\nwants to make money, he is effective in his own way.\n\nI want sympy to be useful for me first, and to protect it from abuse.\nAlso I think in such a case it should be useful for people like me,\ne.g. physicists.\n\nUntill now I think I was effective in makeing SymPy useful, but not\neffective in protecting SymPy from abuse.\n\nAnd now I'm trying.\n\n> > For me I just can't afford to spend that _precious_ spare time without\n> > legally protecting the result from abuse.\n> \n> I think I understand how you feel. But as I said above, I think LGPL\n> maybe protects some abuse in some situations, but definitely not all\n> abuse in all situations (see my example with gmail above) and also\n> from the way you look at it, I don't even thinkl it is abuse --\n> depends on circumstances of course.\n\nThere is no 100% perfect solution, but this does not mean we should not\nat least create some protection. and as i've already said LGPL is a\ncompromise between BSD and GPL.\n\nI've tried to show in every possible detail, that with LGPL in fact\nthere is just *NO* barrier for users, and (to me) very irrational\nbarrier for developers + BSD creats its own barriers, while, as Andy\nsaid LGPL protects the community.\n\n\n\nSummary\n=======\n\nBSD is bad because weird companies can take the code and eat it and give\nnothing back. Big examples are:\n\nMicrosoft, who took BSD IP stack,\nApple who took code from FreeBSD to their OSX\n\nhttp://en.wikipedia.org/wiki/BSD_licenses#Proprietary_software_licenses_compatibility\n\n\nLet's also look at what other developers think:\n\nLinus: “making Linux GPL'd was definitely the best thing I ever did.”\n\n    http://en.wikipedia.org/wiki/History_of_Linux#Linux_under_the_GNU_GPL\n\nFernando: \"I happen to actually really like the LGPL.\"\n\n    http://groups.google.com/group/sympy/msg/5f757166b695d35c\n\nGary:   \"I will never contribut to BSD project\"\n\n    seen on IRC if I recall correctly\n\n\nGael:   \"Companies like BSD because they can keep their modifications\n         secret\"\n\n    http://groups.google.com/group/sympy/msg/74a5c4bc7506ab68\n\n\nPearu:  \"May I propose something that is more relaxed licence model than\n         GPL: LGPL or BSD-like that numpy has?\"\n\n    http://groups.google.com/group/sympy/msg/1a1ee12c8c2928c1\n\nOndrej: \"What I like on GPL is that if someone takes it, he needs to\n         distribute his changes and code also as GPL, and cannot just\n         steal it and used it in a proprietary product without releasing\n         his code as opensource. I have nothing against someone taking\n         our code and selling it, however, I want him to release it as\n         open-source (GPL). I think that's fair.\"\n\n    http://code.google.com/p/sympy/issues/detail?id=138\n\n\n         but then you conclude that \"BSD is better than GPL\"\n\n\nKirill: \"I understand that licensing is only part of the game, and in\n         hard times I put the question on the shelve fighting against\n         the fork and defending SymPy, but now when I'm almost exhausted\n         I just cant afford myself spending _precious_ free time and\n         _health_ on SymPy if it is under BSD license, because companies\n         like M$ or Apple or someone else can take it and make it\n         proprietary.\n\n         I'm 100% ok for someone selling sympy, makeing proprietary\n         applications or other libraries on top of it, but I 100%\n         request for me that sympy itself must be always free as in free\n         speech.\n\n         LGPL is good for both: for me it's sympy protection, for\n         other's it's fair use, even if they want to make their products\n         proprietary.\n\n         It's fair, and it's a win-win for all!\n\n\n\n         Companies who benefit from BSD cultivated that \"BSD is better\n         than GPL (and automatically LGPL)\" mood just to benefit them.\n\n\n         When you are a student with lots of free time and energy, you\n         are \"young and naive,\" and can afford yourself to ignore the\n         issue, but\n\n         I think when it boils down to that you have eps free time to\n         work on the project and you spend your health on it, you have\n         to think very carefully whether you can afford yourself to play\n         this game.\n\n         I still love SymPy very much, but the price to play this game\n         under BSD rules is just too high for me.\"\n\n\n    this mail\n\n\n\nAlso interesting:\n\n    http://en.wikipedia.org/wiki/GNU_Lesser_General_Public_License\n\n    http://www.dwheeler.com/essays/floss-license-slide.html\n    http://www.dwheeler.com/essays/gpl-compatible.html\n    http://www.dwheeler.com/blog/2006/09/01/#gpl-bsd\n\n\nIn Linux kernel case, GPL effectively works like LGPL!\n------------------------------------------------------\n\nThat's because we can think of the kernel like of a libary, to which we\nmake system calls from user space. And since GPL works only at kernel\nspace, user-space code can use whatever license it finds best, so see,\nthis\n\n    user-space  |  kernel-space\n\npartitioning works similarly to\n\n    application |  library,\n\nor\n\n    library2    |  library1\n\npartitioning.\n\nAnd since Linux license don't affect anything beyond that borded I claim\nthat in practice this works *exactly* like LGPL.\n\nAnd now:\n\n> Linus: “making Linux GPL'd was definitely the best thing I ever did.”\n>\n>     http://en.wikipedia.org/wiki/History_of_Linux#Linux_under_the_GNU_GPL\n\nand\n\n>     http://www.dwheeler.com/blog/2006/09/01/#gpl-bsd\n\n\nP.S. in that blog from 2006 dwheeler says that BSD projects can do fine\ntoo, e.g. Apache.\n\nI have news for you that now M$ is actively trying to affect apache\ndevelopment:\n\nhttp://lwn.net/Articles/291628/\n\n\nP.P.S. BSD encourages companies to build on BSD code and hire brightest\ndevelopers, but as described in that blog entry from 2006 it's bad for\nthe project and it's bad for that developers in the long run.\n\nFor some time I started to think most of scipy/numpy/sympy developers\nare positive about BSD because this opens job opportunities for them at\nenthough/etc...\n\nI really hope that Linux example shows that with good managment and\n\"must cooperate\" (i.e. can't keep modifications secret) rules, talaneted\nteam and bright ideas, it should grow up and\n\n    set _our_ rules for companies instead of being driven by potential\n    sponsors wishes and desires.\n\n\nThis all shows that Protected-Freedom and Independence is always good.\n\n\nAnd again, this is *weak* copyleft! Anyone can build proprietary things\non top of it!\n\n\nIf there are people, who share my views, please\n\n __________________\n( Vote for LGPL!!! )\n ------------------\n        o   ^__^\n         o  (oo)\\_______\n            (__)\\       )\\/\\\n                ||----w |\n                ||     ||\n\n\nThanks beforehand,\nKirill.\n\n\n\nAppendix I\n==========\n\nFor the reference here are some excerpts of our internal conversation\nwith Ondrej which I've edited and extended. I only quote myself so I\nthink this is ok.\n\nThe text is a bit sharp -- that because I call things by their names.\n\n\nSymPy was already abused\n------------------------\n\nSymPy was already abused (\"old new core\" and [1]), and if i recall\ncorrectly, it was me, who told you what to do (\"lets do patch review\")\nand defended this and made everything possible to not letting the\nproject split.\n\nNow I'm telling you that when we move forward, sooner or later, there\nwill be another kind of abuse, but you don't want to listen to me. It's\na pity and makes me sad.\n\n[1] http://code.google.com/p/sympy/wiki/SymPyCore\n\nCommunity is not weaker\n-----------------------\n\nThe problem with your approach is that you implicitly assume that\ncommunity is always weakier and companies are always stronger, so it's\nbetter to receive at least anything than nothing.\n\nBut I say, that is should be that every player plays on a common ground\n-- a company, a developer, a community at large, another company -- all\nmust have equal rights with respect to sympy.\n\n\nFreedom have to be protected\n---------------------------\n\n(1)\nAnd when you are only in university environemtn that's ok, but I think I'm not\nthat dense case who got up almost up to phd in university and also to maybe the\nsame level in industry.  And being in both this two places i can say for sure:\n\"freedom have to be protected\" i think people part to those who know this, and\nto those who will know this.\n\n(2)\nMy another topic is that freedom *have* to be protected. BSD was ok in\nuniversities when/where there was already a culture to share scientific\nresults, (at least after they are published).\n\nBut this days we have another situation -- there are a lot of greedy\npeople/companies, when SymPy matures they could take it as a start, hire\nqualified developers, and run it.\n\nWe are hobbyst, it would be impossible to compete with full-time gang.\n\nI hope you at least agree with \"Fair use of SymPy\". If yes, I take the\ntask to find relatevile easy way to make it legal and try to convince\npeople. ok?\n\n\nOn John Hunter's License Pitch\n------------------------------\n(http://www.scipy.org/License_Compatibility)\n\nGrr, I'll sumarize what John pitches about LGPL:\n\n1. many companies are still loath to use it because of their phobias\n2. you cannot reuse LGPL code in a proprietary product\n\nThat's all.\n\nMy answers:\n\n1. If a company does not understand what our license means what could we\n  do? All the world understands it, and if we'd put \"Fair Use of Sympy\"\n  before the license to show our spirit I think it would be clear for\n  99.9% of people.\n\n\n2. You *can* reuse LGPL code in a proprietary product in the form of\n  library -- look at glibc - it is LGPL an every program reuses it. After\n  all good code reuse is not copy-paste, but reuse in form of libraries,\n  so what is the problem here??? (Yes, if you want to modify the library,\n  you have to share your changes, but only to the library, and other your\n  code stays only yours)\n\nWe all BSD, Proprietary, LGPL, GPL, DONTCARE *use* LGPL library - at\nleast glibc (LGPL) and libgcc (GPL + linking exception).\n\nWhat's the problem?????!\n\nLGPL is the golden median between BSD and GPL\n---------------------------------------------\n\nWhen I first knew about Linux and Free Software, it was a \"no question\"\nfor me that it is a good thing, and I just wanted to be part of it. And\nI took it the balanced way - LGPL. For me it is absolutely ok when I use\nLGPL'ed library at work when we ship proprietary software, and I just\ndon't understand why it is bad for others. I take BSD and GPL as\nextremes in this respect, but why, just oh why the balance can't work?\n\nLGPL examples in common use both in open and closed worlds\n----------------------------------------------------------\n\nYes, and I've shown you other examples recently:\n\n - winelib(LGPL) + google picassa (?) for linux (proprietary),\n - gtk(LGPL) + lots of written stuff both open and closed,\n - glibc (some small parts BSD, most LGPL) + everything,\n - libgcc(GPL + linking exception) + everything.\n\n\nOn about most of scientific stuff around Python is BSD\n------------------------------------------------------\n\nYes, it turns out most of scientific code around Python is BSD, and some of\nthe code was _forced_ to change it's license to BSD (IPython was\nessentially _forced_ to go LGPL -> BSD [1])\n\nI claim again, that because it benefits some companies, they created\nsuch a pro-BSD athmosphere and they've reached criticall mass.\n\nBut again I say -- just open your mind and drop this bullshit -- you\nstarted from fair rules:\n\n\"\"\"\nWhat I like on GPL is that if someone takes it, he needs to distribute\nhis\nchanges and code also as GPL, and cannot just steal it and used it in a\nproprietary product without releasing his code as opensource. I have\nnothing against someone taking our code and selling it, however, I want\nhim\nto release it as open-source (GPL). I think that's fair.\n\"\"\"\n\nJust substitude GPL -> LGPL and this is ok for all.\n\nWith LGPL we are _not_ forcing anyone to be of some particular license\n-- software B which uses SymPy can be of esenntially _any_ license both\nproprietary and open.\n\nAll we need then is good team and hard work and if we succeed they will\n_have_ to use us.\n\nBut I really don;t care about whether we are in EPD.\n\nWhat I care about is that SymPy is useful and fast for my physicists use\ncases, and also I care about other guys problems who ask questions in a\nconstructive manner.\n\nAnd I claim that doing so and starting from small set of goals, if we do\neverything right we'll achieve greater goals. Success is _never_ planned\n-- we should just do what is best for us. And it is _completey_ _fair_\nto use some weak protections -- LGPL. By myself and Gary and a lot of\nother developers I can say that they actually _care_ to which project to\ncontribute -- a lot of them tend say \"will never contribute something\nnontrivial to BSD.\"\n\n[1]\nhttp://projects.scipy.org/pipermail/ipython-dev/2004-October/001398.html\n\n\nLGPL says \"must cooperate\"\n--------------------------\n\nLet's ask ourself - to help what?\n\nIf it is just used by Mathematica - that's no problem in any case -- in\nBSD and in LGPL, because we explicitely allow that both in spirit and in\nlegal words.\n\nBut didn't Wolfram ever hit any bug in gmp? Even a small bug?\n\nI doubt it -- no software is perfect.\n\nSo if\n\na) they are hitting a bug, and\nb) they want to ship modified gmp\n\nthey _must_ cooperate on gmp.\n\nEven if they don't send patches to gmp itself, they have to provide\nsources to their users, and soon they should learn that maintaining lots\nof patches could sometimes be problematic.\n\nSo they are essentially _forced_ to fix that bugs and contribute back.\n\nIn this sense LGPL helps, and BSD don't.\n\n\nAnd also, should they wish to enhance gmp non-trivially they will _have_\nto cooperate, because gmp is LGPL. I don't know what the probablility\nfor this is (Wolfram wants to enhance gmp), but I'm sure it is not zero.\n\nAnd voala, all should benefit for _sure_.\n\n\nBSD is in fashion\n-----------------\n\n(1)\nCompanies like Enthough and others who benefit from BSD code tought\nsome developers to \"like\" BSD. Some really believe BSD is better, some\njust want to be in fashion and to have job oportunities.\n\nI want fair use and freedom. (and I'm ok for others to build other\nsoftware components on top of SymPy whatever license they want,\nincluding proprietary), so it's not viral!\n\n(2)\nI don't see much legal problems with LGPL v2 and LGPL v3 -- all problems\ncould be overcomed if one want this thing in spirit, but what I see now\nis that new generation\n\n \"Does not want to care at all\"\n\nAnd all those viral BSD nature only helps it -- everyone wants to be\nBSD, so there are no \"legal obstacles and scary things\" people could get\nafraid of. That's it - most people just don't want to care because they\nwere taught not to, and the only lesson from history is complete\noblivion of all of its lessons, its only ~ 20 year ago, all the world\nwas \"proprietary\"...\n\nKirr wants protection\n---------------------\n\n(1)\nLook at me, Fernando, Gary etc.\n\nInitially I was in a shape and could afford myself working on sympy\nwithout protection, but when it started to be very tough, I have to\neither have 100% protection or stop risking my health. So you see, at\nleast from me and Gary \"BSD is less patches,\" and also Fernando says he\nlikes LGPL very much, so I'm grateful to Pearu that he started his fork\n-- I had to overwork myself and feel very strongly that the issue must\nbe resolved or I have to stop spending my time.\n\nI don't want to force anyone, but also I don't want to force myself to\nplay by the rules I can't accept.\n\n(2)\nBecause I don't want SymPy freedom to depend on us always being good\nmanagers and always being full of time and energy etc.\n\nI want protection and insurance, that in no case it goes proprietary.\n\nWhy? Becasue I've put and I want to put my energy and my soul into the\nproject, but I can't afford myself for SymPy to go proprietary or be\nforked or be used unfair and I want to be 100% sure.\n\nSo not \"why not LGPL,\" but I myself subjectively want a _must_ have\nprotection. For sure, 100%.\n\n(3)\nAs I'm telling you all the time, BSD is ok for proprietary closed forks.\nLGPL protects and enourage to give back.\n\nNow with good management we'll have better chances to run, but even if\nwe fail at management, or will run completely out of time, or someone of\nus is hired by e.g. Enthough,\n\n    SymPy always stays independent and free as in free speech.\n\n\nWith BSD they could hire some key developer(s) for full-time job and\ntell them control SymPy -- _they_ will decide whether SymPy stays open\nor closed. Developers are just \"engenieers\".\n\nI think if I care about the project, I have to protect it from the above\nscenario, and so I think you'd better too.\n\n(4)\nYou ask \"why LGPL?\", the answer is that\n\n LGPL does guarantee that something stays free as in free speech always.\n\nIn contrast BSD does not provide such guaranties - one can jump to\nproprietary at any point.\n\n(5)\nBut doing things right is always harder than flowing along the stream...\n\nUnfortunately to prove myself right I'd need two things:\n\n1. a lot of time passed to see what way SymPy evolves and how it is used\n2. realized probability that SymPy is misused.\n\nI don't have this two, and I admit that good managment could keep BSD\nonly project out of bad guys too, but this managment should be *really*\ngood and have to last forever...\n\n\nKirr thinks the same as DWheeler\n--------------------------------\n\nBut isn't the following just my imagination?\n\nCompany X hires key developer Y.\n\nY develops advanced features that go to X's product only. It takes users\nfrom 'open' version of the project and starting from some time you\neither have to re-implement what Y did in their private branch or\nsomehow join them.\n\nThis is extreme case, but I think not so impossible.\n\n\nWhy companies like BSD and why LGPL should be good for them too\n---------------------------------------------------------------\n\nI know Google prefer BSD or X11 or whatever is equivalent.\n\nThey do so here like Enthough -- to have the ability to get the code at\nany point and close it and make benefits from it.\n\nI'm not saying they will do so, they try to do it socially first -- to\nget momentum and benefit, but when things comes to serious competiotion\nthey will do so, I'm sure.\n\n\nMy example was that LGPL allowed Google perfectly ok to build its\nproprietary product on top of WINE and that's ok for all.\n\nIf SymPy was first developed by some company it was that company choice\nwhich license to use, but when it is developed by individauls I think\n_we_ choose.\n\nAnd you know I advocate for choosing for balanced protection.\n\n\nOn PostgreSQL\n-------------\n\nAs to PostgreSQL - yes, it's BSD, but it was originated first in BSD\nenvironment and now PG core developers team _carefully_ avoid working in\none company -- to keep the balance and independence.\n\nSo in PG case they managed to keep it up and running and be independent,\nbut I think everything would be simpler if it will be LGPL.\n\nToo often I hear from people that they would not contribute to a BSD\nlicensed project exactly because of the reasons I think, and this is not\nonly around SciPy.\n\n\n\n\nEPD can ship LGPL\n-----------------\n\nSo my point is this:\n\na. for immediate users it is equally both good to have either BSD or\n   LGPL\n\nb. but for users starting to be developers, and developers who put their\n   soul into a thing, it is important that the software stays free as in\n   free speech.\n\n\nSo if LGPL is ok for users, and LGPL protects software, what's wrong\nwith it?\n\n\nThey usually just compare BSD with GPL, and don't look at the golden\nmedian. Some becase they don't understand the topic, others because they\nwant to benefit from BSD:\n\n o companies: they can keep control on the code and release BSD+their\n   modifications closed, and tie customers to them\n\n o developers: some of them want to be hired/contracted by such\n   companies, and thus they play by their rules.\n\n\nI think _we_ should set the rules, not some company, and if _we_ manage\nSymPy to be succcessful product, EPD would just be _forced_ to ship it.\n\nIt's just bullshit they can't ship LGPL.\n\n\n--~--~---------~--~----~------------~-------~--~----~\nTo post to this group, send email to sage-devel@googlegroups.com\nTo unsubscribe from this group, send email to sage-devel-unsubscribe@googlegroups.com\nFor more options, visit this group at http://groups.google.com/group/sage-devel\nURLs: http://www.sagemath.org\n-~----------~----~----~----~------~----~------~--~---\n"}]}