search

constantinpape / skeletor

Tools and wrappers for skeletonization.
MIT License
2 stars 0 forks source link

GPL v3+ #1

Closed william-silversmith closed 5 years ago

william-silversmith commented 5 years ago

@constantinpape I very much appreciate that you're using Kimimaro and it's helpful for you! It was released to (hopefully) benefit the public and the field. However, the license on this package is not GPLv3+ compatible. Would you mind updating it? This will help make sure that everyone that contributes to and uses Kimimaro will be working within a durable cooperative free software ecosystem. The MIT license is a free software license, but it does nothing to ensure that the ecosystem around libraries labeled that way stay that way.

If you're unfamiliar with software licenses, I can explain in more detail.

Thanks! :)

constantinpape commented 5 years ago

@william-silversmith Oh, so you take copy-left serious ;). I actually didn't check kimimaro licenses before, should have done it.

Unfortunately I can't really change the license here, because I depend on it in a downstream library that is MIT and I don't want to change it there, for two reasons:

If you insist, I will remove the kimimaro dependency. Let me know what you think.

william-silversmith commented 5 years ago

I would like others to use the software readily, but it is also important to create a dual power system which guarantees the ownership of productive tools to the people. The GPL prevents free riding and secures material freedom for people that might be subjected to an otherwise unfavorable deal. I understand the difficulties this license can introduce, but there is also substantial value on offer here.

While it may be possible to replicate this capability for much less than the following quoted figures, I calculated that for a petascale dataset the computational cost was about 500k USD using one method (other methods may be cheaper, for example, I never experimented with skeletopyze). Kimimaro reduces this to hundreds or thousands of dollars (I'm still working on the storage angle which costs tens of thousands). This efficacy was recently demonstrated in fact on a dense 1/3 PB dataset using only a few tens of thousands of core-hours. Producing this software has been a struggle of love of mine since last July (and of my colleague even earlier than that), so probably hundreds to thousands of person-hours have been invested in it. I'm also happy to continue to provide support and advice.

Nonetheless, if the difficulties with accepting the GPL are insurmountable in your situation, barring exceptional circumstances, I must insist that the license be followed. If the proprietary ecosystem can pluck from the free one at will; only an emaciated conception of and material basis for freedom will result. On some projects I use BSD/MIT/etc and you are welcome to include those in non-free applications.

constantinpape commented 5 years ago

Nonetheless, if the difficulties with accepting the GPL are insurmountable in your situation, barring exceptional circumstances, I must insist that the license be followed.

The difficulties are not insurmountable, yet in the current situation changing the license here and downstream would mean much more effort (also for third-parities) compared to the value of having access to the teasar implementation of kimimaro here. (For my current use-cases, I found thinning based skeletonization to be better suited, but that's a different topic).

If the proprietary ecosystem can pluck from the free one at will

Just to stress this point, none of the downstream dependencies that I mange or that I am aware of are proprietary. Although there is of course no guarantee that this will stay this way.

I must insist that the license be followed.

This is your good right of course; I will remove the kimimaro and edt dependencies. Please allow for a few days to do this. Also, do you consider it to be necessary to rewrite the repositories history or are you fine with me changing master? (The kimimaro dependencies is not part of a release yet.)

william-silversmith commented 5 years ago

Hi Constantine, changing master is fine. Thank you for understanding. The field is starting to produce generally useful engineering artifacts. Things that are useful are eventually claimed for the purposes of exploitation unless vigorously defended. If you ever change your mind or find a smaller project that you can integrate it with, I'm happy to assist you.

On Tue, Jun 18, 2019 at 2:47 PM Constantin Pape notifications@github.com wrote:

Nonetheless, if the difficulties with accepting the GPL are insurmountable in your situation, barring exceptional circumstances, I must insist that the license be followed.

The difficulties are not insurmountable, yet in the current situation changing the license here and downstream would mean much more effort (also for third-parities) compared to the value of having access to the teasar implementation of kimimaro here. (For my current use-cases, I found thinning based skeletonization to be better suited, but that's a different topic).

If the proprietary ecosystem can pluck from the free one at will

Just to stress this point, none of the downstream dependencies that I mange or that I am aware of are proprietary. Although there is of course no guarantee that this will stay this way.

I must insist that the license be followed.

This is your good right of course; I will remove the kimimaro and edt dependencies. Please allow for a few days to do this. Also, do you consider it to be necessary to rewrite the repositories history or are you fine with me changing master? (The kimimaro dependencies is not part of a release yet.)

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/constantinpape/skeletor/issues/1?email_source=notifications&email_token=AATGQSMBXLGXIHWMVQF3MZLP3EUVVA5CNFSM4HY3KUEKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODX7TD7Q#issuecomment-503263742, or mute the thread https://github.com/notifications/unsubscribe-auth/AATGQSIROUFFVEFVCKRYLULP3EUVVANCNFSM4HY3KUEA .

constantinpape commented 5 years ago

Hi Constantine, changing master is fine.

Done.

The field is starting to produce generally useful engineering artifacts. Things that are useful are eventually claimed for the purposes of exploitation unless vigorously defended.

I can see your point and don't disagree. For me the priority is different though, being easy adoption by bio-image-analysts and biologists, for whom software licensing is quite a challenge and I fear GPL would make adoption impossible.

find a smaller project that you can integrate it with, I'm happy to assist you.

I will probably need to revisit skeletonization for larger volumes where thinning on a mip level 2 or 3 (my current strategy) will not scale anymore. I will keep kimimaro and the licensing situation in mind for this.