Permissive licenses, community, and copyleft

Posted Oct 15, 2015 20:45 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
In reply to: Permissive licenses, community, and copyleft by mathstuf
Parent article: Permissive licenses, community, and copyleft

The ability to modify an LGPL-ed library within an application is pretty much useless. Personally, I haven't ever seen this done.

In my opinion, the greatest strength of LGPL is that it forces developers to give back changes to the library itself without propagating to its dependencies. I see this as very fair - kinda like Linus' "tit-for-tat" model.

Permissive licenses, community, and copyleft

Posted Oct 16, 2015 1:52 UTC (Fri) by dvdeug (guest, #10998) [Link] (11 responses)

The ability to replace a LGPL library within a binary is useless? Until CVEs stop coming out on libraries, I think the ability to patch well-known security holes in libraries is pretty useful.

Permissive licenses, community, and copyleft

Posted Oct 16, 2015 4:11 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link] (4 responses)

Yes, it's not that useful in practice. Yet LGPL automatically prohibits distribution for platforms where there's no dynamic linking (iOS, mainly).

Permissive licenses, community, and copyleft

Posted Oct 18, 2015 15:51 UTC (Sun) by krake (guest, #55996) [Link] (3 responses)

There is no such restriction.

Relinking an application is just more complex if the library is not loaded by the runtime linker.

Permissive licenses, community, and copyleft

Posted Oct 19, 2015 0:06 UTC (Mon) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

Yes, there is, in practice. That's why QT still has a costly proprietary license, for example.

Please, educate yourself before posting.

Permissive licenses, community, and copyleft

Posted Oct 20, 2015 5:37 UTC (Tue) by krake (guest, #55996) [Link] (1 responses)

> Please, educate yourself before posting.

Interesting proposal from someone who obviously has not seen the license text of the LGPL.

Spoiler: it does not even contain the string "static".

> That's why QT still has a costly proprietary license, for example.

Some recipients do not like the clauses of the LGPL, e.g. their legal department doesn't want to look into copyleft licenses or they want to incorporate the Qt code into their own directly and not link it.

And there are those who, like yourself, fell for the rumor that statically linking requires the proprietary license.

However, people falling for a rumor, no matter how many, does make the license actually say that.

So, in practise, people think there is a restriction which, in practise and theory, does not exist.

Permissive licenses, community, and copyleft

Posted Oct 29, 2015 17:36 UTC (Thu) by pboddie (guest, #50784) [Link]

Indeed. Apart from easily-spooked lawyers (how will they survive Halloween?) a significant reason for paying for "commercial" (but not necessarily proprietary) licences is actually for the support benefits you get for your money. Which is why something like PyQt is still actively maintained whereas PySide (which started out as Nokia's feeble attempt to capture the market created by PyQt) is a continuous demonstration of the tragedy of the commons.

And, once again, the "commercial" aspect need not have anything to do with proprietary licensing. There are companies making money selling support for Free Software. Maybe even HP manages to do so.

Permissive licenses, community, and copyleft

Posted Oct 16, 2015 6:00 UTC (Fri) by raof (subscriber, #57409) [Link] (5 responses)

> The ability to replace a LGPL library within a binary is useless? Until CVEs stop coming out on libraries, I think the ability to patch well-known security
> holes in libraries is pretty useful.

That's useful, yes, but is the license the best place to enforce this? Are there any other best-practises that should be mandated by the license?

I, too, would like a “LGPL, but usable on Android, iOS, with Rust, etc” license. It is not unreasonable prioritise the “contribute back” aspect of the LGPL over the “user freedom to replace components” aspect. (Nor is it unreasonable to prioritise the latter over the former, although I don't do so).

Permissive licenses, community, and copyleft

Posted Oct 16, 2015 14:58 UTC (Fri) by pbonzini (subscriber, #60935) [Link] (4 responses)

You can always use the GPL and add an exception for linking. It's used by a few FSF projects too (e.g. pre-GPLv3 libgcc).

Permissive licenses, community, and copyleft

Posted Oct 16, 2015 17:54 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link] (3 responses)

I actually like the better technical language in GPLv3, but I'm a bit too scared about the interaction of additional clauses and the license itself. Can a clause override something in the license itself?

Permissive licenses, community, and copyleft

Posted Oct 18, 2015 0:23 UTC (Sun) by Conan_Kudo (subscriber, #103240) [Link] (2 responses)

IANAL, but generally it is accepted that you, as the author, can grant additional freedoms on top of the license you chose. That's why the FSF has used "linking exceptions" in some of their projects, and other projects (like OpenSSL) have done the same.

In virtually all countries, copyright rules are largely in favor of the author. The only thing an author can't do is say "license A + license B are compatible" without modification if the rules of the licenses are written to not be so. So essentially, you can't be delusional.

Permissive licenses, community, and copyleft

Posted Oct 18, 2015 2:48 UTC (Sun) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

It's one thing is an exception clarifies some license clauuse (like: "We don't consider static linking to produce a derived work"), but if your exemption clearly contradicts another license clause then situation becomes more interesting. I wouldn't risk it, personally.

Permissive licenses, community, and copyleft

Posted Oct 22, 2015 8:20 UTC (Thu) by Wol (subscriber, #4433) [Link]

Personally, I'd say that if X licences their code under licence Y, but says that licence Z is compatible, then that means that in case of argument licence Z wins.

X has effectively said "you can licence my code under Z", but if the people who own other code Z don't want it licenced under Y, then it isn't available under Y.

Cheers,
Wol