Licensing rant / debate thread

[astyle]

I'm not talking about GPL violations because I consider those like thefts. I'm more curious about BSD licensed software and tools, for which companies have no obligations at all and thus all they give back is only due to their willingness to do it (regardless of their motivations).

Try reading the "Donor List" that the FreeBSD foundation publishes.

Ok, this isn't going the way I thought it would. Too political for my taste, sorry to have bothered you.

Well, if you dig deep enough into how licenses even work, and the reasoning for different licensing terms to exist, you do eventually get to politics and opinion camps. The point of my thread was really to talk about why software licensing even matters. Reason being - Yeah, there's this whole unpleasant, political morass of rules and limitations on what people can and cannot do with their own devices that run software. Yeah, it's kind of important to know how things work so that you avoid stepping on other people's toes.

But - there does seem to be a lack of understanding when software licensing matters, and when it really doesn't, and to what extent. This thread is kind of an attempt to crystallize the distinction between such situations.

 

[Cath O'Deray]

… the usual names are Netflix, NetApp, Juniper, Sony, Apple at one time and so on. …

When I counted things for another organisation in April:

"two thousand, four hundred and eighty-six sponsored commits".

More broadly (two sponsors, three trees):

git -C /usr/doc log --oneline --no-expand-tabs --extended-regexp --grep='Sponsored by:[[:cntrl:] ]{1,}Netgate|Rubicon'

git -C /usr/ports log --oneline --no-expand-tabs --extended-regexp --grep='Sponsored by:[[:cntrl:] ]{1,}Netgate|Rubicon'

git -C /usr/src log --oneline --no-expand-tabs --extended-regexp --grep='Sponsored by:[[:cntrl:] ]{1,}Netgate|Rubicon'

src tree, Klara or Netflix:

git -C /usr/src log --oneline --no-expand-tabs --extended-regexp --grep='Sponsored by:[[:cntrl:] ]{1,}Klara|Netflix'

And so on.

 

[sidetone]

As an alternative to Open Source, there's also Fair Source.

• https://fair.io/
• https://github.com/fairsource/fair.io

It includes the non-compete licenses of FSL and BUSL (also known as BSL). Non-compete licenses are where code converts to true opensource after a set period of time, in order for the author to not give an advantage to competitors over their own work, but for them to share it to open source eventually. FSL may be a more practical license than BSL, but BSL was an example template for FSL.

Learned about the term Fair Source from https://techcrunch.com/2024/09/22/s...-avoid-the-pitfalls-of-open-source-licensing/.

Edits: There's also Fair Core License (FCL), which is derived from FSL, that has added protections for specific needs.

Also, I think that Fair in Fair Source means, fair to companies which are the original authors. That's good though, because companies have a business model, and are there to stay in business, then through standardized non-compete licenses, they eventually give back to opensource. It's also good that companies get together to promote these license models.

Fair Source has a requirement that, code have delayed opensource publication, which is a part of noncompete licenses. Fair Source used to have another meaning, which was a more complex and restrictive for use type of license. The stewards of that willingly transferred the domain and other IP to the current Fair Source project, which started in 2024. That previous license project has been defunct since 2016.

I wish I could see the same support for CDDL1.1, that companies get behind it, as upholding a standard of a type of Open Source. CDDL1.1 isn't included as an accepted Open Source Initiative standard, even though it's a good license for businesses. CDDL 1.1 also doesn't seem to be a Fair Source type of license, even though it protected business IP. I like that CDDL protects code from being absorbed into viral licenses, protects business IP of shared code, and it allows companies to share it. There's no reason, that CDDL 1.1 can't be an extended family of Fair Source, or that there be a Fair source license that transfers the license to CDDL 1.1 after a set amount of time.

 

[Ferocious1881]

Neither Photoshop nor Gimp would arbitrarily shut down just because you're using the software for legally (or even ethically) questionable projects...

It should not be the inverse?

I think it unlikely you could successfully do that. You might if you showed a very high level of negligence, but how many precedents (and across how many jurisdictions?) are there of people suing successfully for consequential damages from software bugs. Most licences will contain clauses that shield developers from liability.

In any reasonable jurisdiction, renouncing from reasonable liability should requires scary warnings. Otherwise contract shall be null.

The EU's Cyber Resilience Act will make such clauses entirely invalid for some classes of software.

Which is an awfully malicious idea, it extends and ratify the goverment as the ruler of your personal choices and what you you value and not value, in any case, one can just block the European Market, and put one's headquarters in Argentina.
Also, this law only will be bad for the customers of Europe, that will get only higher prices for nothing in return.

Licenses can be seen as a tangible guarantee of quality.

Nonsense, that goes into reputation and the rent/sell contract, if there is not contract, in any reasonable jurisdiction, it means not guarantees from the software vendor, as it is distributed as is, not guarantees. Thought the software license may be seen as part of the contract in proprietary products, when is a software with an openly available source code, as I thereby declare open source means, I find it as no holding itself as a valid contract.

For private end users, license terms do have the effect of taking options off the table (Like not wanting to pay for Photoshop or AutoCAD, leaving you with GIMP and Blender). For those who are NOT private end users (think employers and distributors), it's more complicated, and that's where it pays to know what the license terms are, what is negotiable, what's not, the risks, the differences in available options, and other related details.

* you think corporations are evil, and only will give back if there is a legal lever which requires it? Use GPL.

If you not want nobody to use your software for profit, do not share it.

We, Berkeley, have a complete and well fitting OS that runs perfectly fine. Linix, OTOH, have just a kernel and a bunch of assorted stuff. Nevertheless Linux got extremely popular. If there were no Linux, would our OS be similarly popular? No, because shops would just cannibalize it and sell it under their own name. With Linux they cannot do that, they must provide it as Linux. And as a consequence, everybody today knows Linux.

Hard disagree, and by the way there is not canibalization.

I would write an essay about how much I despise viral licenses, and consider the GPL as a demonic licensing model, but will reserve to call out a case of extraordinary nonsense.

Imagine Immogen lives in a city with a frequently traversed park in the center of the city, which has even foreign travelers, and one day Immogen for some arbitrary reason or emotion decide to put a table with fruits in the park to everyone every day and put a sign saying, "you can take what they want."

And everyday Immogen repeats the practice, every day for years, years in which foreigners with thousands of gold ingots, local entrepreneurs, starving children, no extraordinary people and beyond the imagination, had taken and used the fruits to their benefit, allowed orphans to become successful and all more.

But a day Immogen feels stole of how much money had all those successful orphans earn, of how much they had archived, and the few cents that the rich had save in fruit, and how nothing she got, thus decides to complain and harass anyone that had even taken of her table and had not given her a quote, in all those years of giving it for free.

The morality of Immogen is the same that the proponents of things like Open Zero, as well the morality of the bunch of proponents of the so called open source movement, the claims of maintainers that complain about "not remuneration", proponents of the ideology of the FSF tend to at least share in a personal level.

 

[sidetone]

I've given it more thought. If going for practical terms, and not on widespread compatibility with overly viral licenses which overextend into libraries such as GPL, then Apache 2.0 is perfect for libraries. A copy left file or directory based version of Apache 2.0 would be good too. MPL2.0 is close to that, except it allows the license to be upgraded according to Mozilla's discretion. Still a very good license. Then, CDDL1.1 is great for endpoint software products, which use those libraries.

CDDL1.1 and Apache 2.0 are beneficial due to their additional protection of patent use. Permissive licenses are good too, for major software organizations which are big enough which can protect and maintain them. So, Apache 2.0 and CDDL1.1 would be good for the ports tree, and for smaller groups of authors

Then, GPL allows use with a Classpath exception, which when this is added, other code is allowed to use GPL code as libraries without extending the viralness into that code. GPL doesn't need to extend into libraries, which are dynamically linked, period. GPL2 is already incompatible with LGPL 3 versions due to this.

Also, in the FreeBSD ports tree, many Apache licensed programs are often missing components, or rely on GPL configurations and GPL build tools. Apache licensed programs are still free to be made in a BSD way, or at least a non Linuxism way as well.

If an organization or for-profit wants to offer a product, then CDDL1.1 is more straightforward for users for open source than GPL.

Looking the "Complete Guide for Open Source Licenses" PDF from MEND, it appears that even the Microsoft Public License is better than the GPL. It says, it is incompatible with it, due to that it doesn't want code being given up in a "blackhole" to owners of GPL code. That the MS-PL license is less restrictive in forfeiting code than GPL is a major irony, considering Microsoft's history and Bill Gates unfavorable interactions towards Linus.

Back to, about why CDDL 1.1 isn't an accepted license at the OSI. https://lists.opensource.org/pipermail/license-review_lists.opensource.org/2024-July/005502.html says, that it may have not been submitted, and if it were, it would be so for legacy purposes. That very few programs used this license, and most code was under CDDL 1.0. Even so, I believe it would be good for the CDDL 1.1 license to be submitted or replaced with a similar license anyway. Even if limited programs use it, it would still be great as being Stewarded by the license creator. It's a good license, and it didn't get the popularity it deserved, as it was overshadowed by GPL's rhetoric.

Edit: The clause specific to California Law added in CDDL 1.1 also makes it, not a universal license. That may be applicable to where Oracle is located. Still, it should be submitted to OSI. Then, we need a license like it on the basis of that acceptance.

Also, that there isn't a popular universal license version of CDDL 1.1, means the opensource ecosystem is severely lacking. MPL 2.0 is great for open source, however, it serves a different philosophy of being upgradable to a new version. GPL dictated what opensource should be, when they should just make a universal dynamic linking exception to work with other libraries without absorbing them. This level of viralness is what's wrong with GPL.

 

[hasbeen]

I had a dream about "Open Internet and FOSS". Tho, that was just a dream.
Then, I thought what would Internet look like today without monetization of the Internet and almost everything on Internet, and who would be concerned the most about Open Source Software licensing schemes. That aside, here is interesting take on Linux:

View: https://www.youtube.com/watch?v=eqtN0lgzabE

 

[6502]

I had a dream about "Open Internet and FOSS". Tho, that was just a dream.

You have to read Karl Marx and his dreams.

 

[astyle]

I had a dream about "Open Internet and FOSS". Tho, that was just a dream.
Then, I thought what would Internet look like today without monetization of the Internet and almost everything on Internet, and who would be concerned the most about Open Source Software licensing schemes. That aside, here is interesting take on Linux:

View: https://www.youtube.com/watch?v=eqtN0lgzabE

This is off-topic, unless you can connect this to the topic of this thread, in a manner that other participants can understand.

If you want to make a connection between A: license terms and B: global politics, please expand and explain how they are connected. Otherwise, spreading of unsubstantiated FUD like this is generally discouraged on these Forums. Even Youtube is officially making an effort to tell people to avoid off-the-cuff remarks and think before you post...

 

[masc.h]

Great thread! Thanks for referencing it astyle.

Another classic example of why licences do or do not improve a project is NGINX, written by Igor Sysoev and licensed under the BSD 2 clause.

If we research a little, we figure out fast that most of the web is powered by nginx, open source and "cuck licensed". Still, it did not stop F5 Networks nor Sysoev's nginx plus. Quiet the opposite developed, million to billion dollar of revenue. These projects advance. You appreciate those people who found a solution and committed it to the public for almost nothing. They're simple people with passion. GPLs and Linus Torvals are delusional, they follow an ideology nothing more.

Listen, many of us probably work from 7am to 3pm or even 4pm. We all do our work as good as we can, we try to be productive, passionate and come up with solutions and still our loan stays the same, but between our colleagues we gained reputation and our mind feels great, we achieved something. Such technology passion is like the passion kids have when building castles of sand, ... most do it because they simple live, out of fun, adventurer instincts, ... They dont expect a billion dollar back. I havent seen Sony or Apple talk big about their kernel and operating system, being insulting or anything like that... Linus on the other hand said himself, he likes the feeling of being credited with other peoples work. Honeslty Im not mad for paying 500€ for a Playstation and enjoy the awesome game and graphic experience. Until i got such availability in Linux or got it configured to work properly and fine, I bet with you, by that time you could have bought 3xPlaystations.

At the bottom line, GPL should let go. All those ideologists. They try to enforce a memorial and being credited. On rough streets that wont work for example.

 

[astyle]

Well, I did make a profile post, but it probably really should be in this thread...

I gotta say, RMS makes some good points, but at some points, he stops being academic in his explanations. As an example, RMS deliberately mangles DRM to mean 'Digital Restrictions Management', and uses casual language to mock the very existence of it. (Re-read the section titled "Powerfu, Reliable Software Can Be Bad"). The way I would summarize that section is: "Even if source code for the Digital Rights Management software is publicly available for inspection and free usage, DRM is nonetheless something that GNU philosophically opposes".

Considering what DRM even is, it's not that hard for a sufficiently talented programmer to implement a pretty complete commercial-grade DRM system using all-GNU toolchain and components. GCC, GNUPG, Guile, and more. But - DRM is becoming a rather pointless thing - even Apple stopped tagging all of its iTunes music with DRM after 2008. RMS' explanation is from 2013.

 

[kpedersen]

even Apple stopped tagging all of its iTunes music with DRM after 2008. RMS' explanation is from 2013.

True but unfortunately added it to the aarch64 specific builds of the macOS installer. Example 1, 2, 3. They also kept the stupid Windows RT style "developer license" to run code on your own iPhone.

DRM is here to stay. Its not about money anymore. It is the control that the vendors want.

And Red Hat is a good example of how a completely open-source ecosystem can deploy DRM. For example their packages are effectively locked behind a DRM mechanism.

 

[astyle]

DRM is here to stay. Its not about money anymore. It is the control that the vendors want.

I think that when it stops being about money, control mechanisms like DRM have less leverage in the market, against either circumvention or or non-DRM alternatives.

Controlling a production pipeline in the name of quality product is one thing. Controlling a market to maximize your own benefit is not nearly as easy.

 

[sidetone]

Proposals:
LGPL substitutes
A license which includes Apache 2.0 license, and waives the patent clause only when used in combination with GPL. An LGPL substitute to replace LGPL2, with added patent protections when not used with GPL, and for use with all GPL versions. Use mainly for libraries. This is permissive, while LGPL is file-based copyleft. Making this file based may add complications of marking the license to it using Apache 2.0. APL2.0 code is protected by use of requiring marking changes to code. For a library, file based copyleft isn't required. By allowing its use with future GPL license versions, this license could inadvertently allow those GPL versions to eat into the license, by this modified APL version allowing the patent clause to be waived for use. Maybe, allow future discretions of which patent clauses to waive in the license update for future versions of GPL. The only part of this license that could be updated, is the patent clause waiver for future GPL versions, and be done by stewarding organization or by author.

Alternatively, license which includes LGPL2.0 with additional patent clause which is waived when used with GPL3, GPL2 and any future GPL version. This would be file based route, instead of permissive route. For use with GPL, using an LGPL base may be safer, as to be compatible with new future versions of GPL, as it could find a way to eat into the patent clause of the APL 2.0 license, by the modified APL license allowing the waived patent part of its license.

GPL substitute
Includes GPL license, where GPL doesn't extend into code used as libraries for it which are dynamically linked. Major and minor versions match that of the GPL version it's used with or amended to. This functions as a link exception, except that it's clearly defined to downward used libraries. Link exceptions upwards can still exist with this.

File-based directory wide license
Similar to Apache 2.0, except file based directory wide license. Contains patent clause. Simple, as follows permissive license ideology, but doesn't allow dynamic linking restrictions simply based on use with other open source licenses.

Some language in Apache 2.0 isn't needed. It can be worded more like BSD or MIT licenses, with few important additions: file based directory wide, patent clause and doesn't allow use with licenses which restrict dynamic linking to it. Patent clause could possibly be waived for when use with certain licenses, but since this is GPL incompatible, this isn't necessary. Mainly not a widespread library license.


Rationale
The GPL alternatives are the simplest way to allow wider compatibility use of libraries through dynamic linking. It reduced viralness, amends GPL licenses so they can be used with varying LGPL, APL, CDDL and MPL libraries. Authors agreeing to amend to either the LGPL or GPL alternatives each allows that code to be used with more libraries. It simplifies the text of link exceptions, where this often isn't listed in the place of common description. See how CUPS uses Apache 2.0 with link exceptions for use with GPL2. It also reduces the need to combo license used dynamically linked dependencies for each GPL version. This is the simplest way to clean up GPL madness, with the fewest steps, if the stewards of GPL refuse to use this route to make dynamically linked libraries simple to use. As with GPL everything else has to dance around it, and this would be the simplest ways to include it.

The file based directory file based license is simply a GPL alternative, which is meant for use with LGPL and other libraries. It's not meant for use in conjunction with GPL. It's an MPL and CDDL alternative as well, which is meant to be used side by side with them and with other non viral licenses.

 

[rbranco]

To me this whole thing is a mess.

I'm sure there are a lot of software out there, usually packaged as containers, shipped with all their dependencies with incompatible licenses. The nature of containers and interpreted languages makes the dynamic linking point mostly moot, I think.

The Apache License 2.0 is the only Apache license compatible with GPL, and only GPLv3, which was considered extreme by Linus Torvalds.

OpenBSD doesn't include Apache or CDDL stuff.

I stick to only BSD & MIT in my stuff as I don't have time to read more than one page of legalese.

 

[astyle]

OpenBSD doesn't include Apache or CDDL stuff.

yeah, but it does have Apache in its ports, and doesn't give a rat's ass about that.

I stick to only BSD & MIT in my stuff as I don't have time to read more than one page of legalese.

Point of this thread is to explore when a license even matters, and why. Sometimes, one just needs a certain feature to get the job done, and that really trumps a given license's terms. I want my copy of software on my machine to work correctly for me, regardless of what license it has.

 

[rbranco]

yeah, but it does have Apache in its ports, and doesn't give a rat's ass about that.


Point of this thread is to explore when a license even matters, and why. Sometimes, one just needs a certain feature to get the job done, and that really trumps a given license's terms. I want my copy of software on my machine to work correctly for me, regardless of what license it has.

OpenBSD: Copyright Policy

I mean in the stuff I write. I also don't give a rat's ass about licenses in stuff I use. I prefer GCC over Clang/LLVM, for example.

 

[astyle]

OpenBSD: Copyright Policy

I mean in the stuff I write. I also don't give a rat's ass about licenses in stuff I use. I prefer GCC over Clang/LLVM, for example.

I stick to only BSD & MIT in my stuff as I don't have time to read more than one page of legalese.

Yeah, it's kind of important to be familiar with the license you wanna use on your own code, but that is very different from being aware of licenses on stuff you use... I'd suggest not confounding the two, especially in the same sentence

 

[rbranco]

Yeah, it's kind of important to be familiar with the license you wanna use on your own code, but that is very different from being aware of licenses on stuff you use... I'd suggest not confounding the two, especially in the same sentence

Who cares about licenses for stuff they use? You're just trolling and wasting my time.

 

[mer]

Who cares about licenses for stuff they use? You're just trolling and wasting my time.

If you are a business, especially a small business that writes software in order to buy food, you and your employees most certainly should care about the licenses for stuff they use.

 

[astyle]

Who cares about licenses for stuff they use? You're just trolling and wasting my time.

The very point of this thread was to explore that very question...

 

[kpedersen]

Who cares about licenses for stuff they use? You're just trolling and wasting my time.

As a user, I really don't care. I know some people get religious about these things. So long as I can access the source, I am happy. And even if not, so long as it doesn't have DRM, I will still use it.

But for many software houses, it is crucial that licenses are kept track of. If you are a small startup, its the first thing an investor will ask you about. If you are a large corporation, a failed audit may cost a serious amount of money which could have been easily avoided.

 

[sidetone]

I was exploring the best way between the viral GPL, permissive licenses, and a simpler way than MPL. Lots of licenses work together well, but GPL's and LGPL's can't always work with each other. I wrote a blueprint for amending GPL, by either them doing it, which they likely will refuse in the immediate future, or simply making a license, which includes the GPL. I like permissive licenses, but they allow themselves to get eaten up, for instance by the GPL or even proprietary code. I also like the idea of using patent clauses, where they can be waived in certain conditions to allow use with other opensource licenses.

The above, and previous posts of mine, were, how can you make GPL compatible for use with libraries used as dynamically linked dependencies. Modified GPL, (with the following unmodified as is) MPL, Apache, CDDL etc should be able to be used together, dynamically linked, for use as libraries in even the minimalist of ways, by keeping separation of code. They already use link exceptions, but I wanted something clearer and more standardized as a license. Instead of understanding every line of the GPL, to simply include which GPL version in the license, and the additional permissions, and where the patent clause is added and waived for use with GPL.

If this gets done, there can be the perfect opensource world, where modified GPL can be used, and doesn't have to eat everything up as a library.

As for the complexity of LGPL, MPL, and Apache, I let the stewards do that, and for the user to understand enough of it. Of these, APL2.0 isn't complex at all, it just needs to be included under another license for the purpose of making it compatible with GPL2, and to make a patent clause which is waived for use with GPL3.

Another core belief of mine is, that people or companies should be able to use a modified GPL, which where they know where its limits extend, so they can use it, without losing code they don't intend to lose. Of course than can and should contribute to it, but knowing those limits allows them to choose where to contribute, and what they wish to keep.

 

[mer]

As a user, I really don't care. I know some people get religious about these things. So long as I can access the source, I am happy. And even if not, so long as it doesn't have DRM, I will still use it.

But for many software houses, it is crucial that licenses are kept track of. If you are a small startup, its the first thing an investor will ask you about. If you are a large corporation, a failed audit may cost a serious amount of money which could have been easily avoided.

I agree that as a user one probably doesn't need to care, but sometimes a license will say "free for personal use, but not for commercial". That can affect what tools are used in commercial settings.

 

[sidetone]

There's a flaw in the LGPL as well, which makes it suitable for libraries, and not end user programs. An LGPL end user program cannot be used on top of another copyleft licensed code through dynamic linking. It has to be locked in to the same LGPL major version of permissive license. But, the good part is that LGPL code can be used side by side to fill dependencies of a program that relies on both. Therefore, LGPL only makes sense for libraries. Though, it's necessary for LGPL to not allow it to use through dynamically linking non LGPL or permissive, so that GPL can still use it.

Back to the GPL, it forces other programs which intend to be for use of all opensource to combo license. Then, when it combo licenses as LGPL2 and Apache together, it wants to force LGPL2 to become GPL, which restricts that code for use as intended by LGPL. Or code from Apache 2 is restricted in the combination of other code in that library added as GPL to that library, once it's incorporated into GPL3.

GPL2 and GPL3 need to allow the use of dynamically linked liberties below them, so those software programs don't have to combo license, and to improve license compatibility. GPL2 should be able to use LGPL3. LGPL, BSD, MIT, Apache and MPL should be able to be used as dependencies of GPL2 and GPL3 without forcing those to be combo licenses.

When code is meant as Apache 2, GPL forces them to be under MIT which offers less protection, or to LPGL2, which lacks a retaliatory patent clause. Apache 2 with GPL2 is an ok combination for allowing use, except, it's complex to require stewarding code encouraging contributors to code under both licenses to be accepted to that project. LGPL also is required to complicate itself further that GPL demands it turn into GPL.

There needs to be an open source standard that states, all programs must allow dynamic linking from libraries. Then you could have licenses meant only for libraries, which are compatible for use, by any that allows dynamic linking from. GPL2 and GPL 3 need an update. This is probably lack of foresight.

Plus, GPL2 needs to be able to use LGPL3 libraries. That's the way to do that, and solve other problems. For the benefit of other software stewards and programs, it decreases complexity of having to multi license to allow compatibility.

It would benefit both GPL major versions to allow these to be used as dynamically linked. GPL2.2 would be allowed to use LGPL3 through dynamic linking. Other licenses wouldn't have to be dual licensed to be used by either version of GPL, which benefits programs which their authors choose a version, but must additionally add it to LGPL2 to make it compatible with both major GPL versions.

Ideally, LGPL2 should be able to use LGPL3 as dynamically linked dependencies, however, LGPL2 is good enough, and it would only complicate it for programs under the existing GPL2. LGPL2 and LGPL3 should also be able to use code that's dynamically linked which is under other copyleft licensed libraries which allow their code to be used for such purposes.

Allowing this use of opensource would is optimal opensource, and how it needs to be. It should be the minimum definition of opensource.

Anything with GPL2 or greater would immediately benefit. Think about the programs which are under an GPL2 with few authors or stewards, they would see the benefit and adapt to this license. Many programs will be left behind, but they'll still be useful in their own ecosystem. Think about if there's an OS under GPL2 (I can think of 2), and the steward sees the benefit, then it can then select from more drivers under Apache 2, CDDL, MPL, LGPL3 etc... without them having to combo license. Then, those drivers also be available for other operating systems without forcing that code to become GPL. The point is flexibility of dependencies and a software which uses them. I believe drivers should at least be under LGPL which has maximum compatibility to be used by other licenses.

A danger of LGPL, however is, that it can convert into GPL. With GPL, it's said that there's a risk of losing all code connected to what was originally meant to be shared. So, you'd have to dynamically link to LGPL to be safe, and only contribute to LGPL what you wish to give away. The GPL doesn't make sense in this capacity, as what one contributes to a BSD license is limited to that, then GPL can absorb code under that BSD license. I'd make a new library opensource code, which is filebased copyleft, allows dynamic linking in both directions, absorbs only what's used with it not through dynamic linking, and can't be converted into another license.

Something that I may not be understanding is, for Apache code, if something doesn't belong in it, the Apache Foundation makes it get out of the code. It's permissive anyway, so a piece of code side by side, as long as it's commented from the Apache code, is fine to use. GPL seems that, if a company puts a piece of code into it, GPL will eat up everything else the company made not intended for GPL simply because it uses a small piece of code it mistakenly put under the GPL. That may be a misconception, but that's how it seems.

 

[sidetone]

Peter Saint-Andre who came up with the idea for Jabber/XMPP, and Paul Edwards behind PDOS (Public Domain OS) have made arguments in favor of Public Domain.

Edwards argued that code should go into public domain, that it makes it free for all to use, and software doesn't have to be rewritten, sometimes excessively, to achieve the same function. That business should have rights to use software as well.

PDOS / Git Code / [4588fd] /pdpgoal.txt

sourceforge.net

Saint-Andre questioned that what's to keep from others use, by not using public domain. https://stpeter.im/writings/essays/publicdomain.html also. https://stpeter.im/writings/essays/whatiscopyright.html

That was about their arguments for Public Domain. Who's Afraid of the Public Domain? I'm concerned about Public Domain or permissive code being gobbled up and made into derivatives preventing their use with other opensource.

MIT license is the next closest thing. Then, Simplified BSD licenses, and ISCL are the next thing, with similar rights for use. Who doesn't want to get credited for their work? As far as I'm concerned, these work similarly enough for allowance of use.

BSD licenses with advertising clauses are fine too. There's no restriction for use, but may be inconvenient.

Apache 2.0 is good in principle. It requires keeping code outside of the license to be commented as not part of the license. I question when Apache 2.0 is used with GPLv3. It seems as if GPLv3 doesn't disclose the parts of Apache 2.0 that require separation through commenting, from GPLv3 which incorporates all with touches its path. Apache 2.0 and GPLv3 are 2 separate completely different entities, and thus not the same license. GPL needs to be required to disclose the boundaries of where Apache 2.0 is within its GPL codebase. I could be wrong, however, it seems this is the case, considering GPL argues for the MIT license that it allows it to be used without disclosing that code was published under license. Though, GPLv3 allows use of Apache2.0, it can use it as linked though separate codebase. However, the GPL argument, is that it allows use of permissive licensing, that it must be allowed to gobble up that code at any time, thus my concern about whether GPL gobbles up Apache without respecting its boundary documentation still stands.

What's ironic is, that GPL closes off opensource more than proprietary use of opensource does. When a company uses permissive or file-based opensource, it happens quietly. When GPL absorbs it, it absorbs it like a sponge, it puts out code for public use, then uses that to keep people in that ecosystem shut off from the rest of opensource. GPL adds derivatives of permissive code, then those derivatives are shut out from the rest of the opensource world. Also, Ironically, is that GPL argues for the use of companies, which, GPL is not a company friendly model. GPL also isn't a friendly model to other opensource, for use. Like Edward's argument, a piece of code has to be rewritten to make it work with different licenses. Different pieces of code for the same function is redundant.

It's ironic that there's the argument for freedom, which permissive licenses allow, then they allow that same code to be eaten up by restrictive opensource through that freedom. Those are the licensing and restrictions by those licenses they argue against. Though, public domain does in a small way reduce the ability for more restrictive licensing to eat up common code.

So, permissive licensing works when an individual, group or organization is enough to sustain a particular code base. Individual projects for instance are enough to maintain code, to keep it relevant, so that there's no purpose for there to be more restrictive versions. Larger pieces of work are able to maintain their code, these require backing institutions. For instance, PostgreSQL, FreeBSD, NetBSD, OpenBSD, Hardened BSD. Ones with foundations have even more leverage to maintain their code. Other organizations also are able to maintain their code permissively to different extents.

The less potent an entity is to carry a project, that file-based licensing is needed to protect that code. In my view, freedom of the use of code with reasonable licenses is important, while at the same protecting code from incorporation into viral opensource.

At minimum, GPL needs to allow it to use libraries through dynamic linking. It doesn't, so I believe the definition of Open Source needs to change to must allow dynamic linking to and/or from. Any code which doesn't allow dynamic linking to and/or from isn't Open Source. GPL restricts itself even from using libraries licensed from its own stewardship. LGPL can meet this minimum, because it allows dynamic linking from. LGPL may not be optimal opensource, but I give it that.

It's like as if Stallman is doing the same, which he complained about. He complained about contributing to opensource, and not being able to use derivatives of his own work, so he formed the GPL. Now, GPL closes off code to other opensource, while a company is able to dual license proprietary with GPL. That company gets to use that code under their terms, while other companies must forfeit their contributions. If you don't allow dynamic linking to and/or from, you're not truly opensource. Any license for an end user program which doesn't realize the need for a new version to allow it to use dynamically linked libraries needs to be replaced. An end user program is intended to use dynamically linked libraries without imposing its own restrictions. If not, it's not truly opensource. The realization of the opensoftware community needs to see that. By definition it may be labeled as opensource, but in my eyes, it's not.

An opensource library is intended to be used by other opensource and perhaps proprietary, while protecting itself. At minimum, an opensource library must allow dynamic linking from. Optimally, an opensource library allows dynamic linking to it. There's no useful purpose for restricting dynamic linking to or from a library. It's an opensource library for crying out loud. Code which allows dynamic linking both to and from, can optionally be a user end program as well as a library. If a library is written in GPL, what's the point, it belongs in the dumpster. If GPL2 and/or 3 are restricted from using libraries under LGPL3, Apache 2.0, MPL, CDDL and other file-based copyleft, it's time to upgrade those GPL licenses to allow dynamic linking to, or replace them, leaving them in the rearview.