This is not legal advice, just a wild-assed estimate of a lawyer thrown 
out without having done any substantive research:

Almost any license not expressly made irrevocable or associated with 
the transfer of a property interest is revocable.  A free license, even 
moreso.  Not that the defendant would not have a litigable issue -- one 
argument against revocation is that the license induced a reasonable 
and material change of position on the part of the licensee.  I can 
tell you that it would be highly fact-specific, and an expensive 
lawsuit to settle the question.  My guess is that virtually every 
open-source license is likely revocable to a point -- until substantial 
material dollars are invested or an express statement that the license 
is irrevocable is made by the licensor.  The Apache license is probably 
NOT revocable, but Squeak-L might be.

On Wednesday, March 26, 2003, at 07:55 AM, goran.hultgren at bluefish.se 
wrote:

> Cees de Groot <cg at cdegroot.com> wrote:
>> On Wed, 2003-03-26 at 08:51, Stephane Ducasse wrote:
>>> Could a group of people identify a path or multiple ones and make 
>>> a=20
>>> clear analysis?
>>
>> Squeak's license is irrevocable(*). So there is no risk at all that
>
> How sure are you of that? I have been trying to find info on
> irrevocability but had a hard time finding stuff.
>
> [SNIP]
>> There's a second argument that's not related to this open source
>> business: Squeak (whatever that means) is distributed under the 
>> SqueakL.
>> So, by reading the SqueakL, you now your rights (use it for whatever 
>> you
>> like, don't sue us) and obligations (share some modifications, don't
>> give it to people living in Libya). If Squeak, as you get it, would
>> contain code under various licenses, you'd need to study all the
>> licenses and it would make it less easy for you to evalutate whether
>> Squeak is usable by you.
>
> Right. And frankly, this is perhaps my main reason for keeping some 
> form
> of core/base (whichever suits your vocabulary) under Squeak-L *only*.
>
> Again, note that other "groupings" of Squeak stuff (think distros or
> whatever) can contain multiple licenses. But it still is nice if there
> is one subset of all possible Squeak packages that can be had under ONE
> license. And the community tendered official subset of those would be
> what we call the core.
>
> (since I don't think we will ever get around to changing it)
>
>> A third argument is that some licenses have viral qualities - if you
>> combine code under two licenses, one license may state that 'the other
>> code' comes under its terms. The GPL is a clear example of this: if 
>> you
>> link any code to GPL-licensed code, the GPL states that the resulting
>> whole counts as a derived work and therefore must be licensed under 
>> the
>> GPL. This is good if your political agenda is called 'copyleft' - in
>> effect, they are founding a protected commons - but bad if your agenda
>> is 'I do not care about political agendas'. Other licenses may have
>> similar terms; for example, the SqueakL stating that everything you
>> modify of Squeak (as you received it) should be shared with the
>> community is viral: if you file-in code that has modifications to, 
>> say,
>> Object>>asString, the modification *must* be shared with the community
>> under the SqueakL (only if you redistribute or sublicense it - read 
>> the
>> SqueakL for the gory details).
>
> Eh, IIRC Andrew even got to the conclusion that you *must* share it 
> even
> if you are not distributing it! Probably an unintended flaw in the
> license, but anyway.
>
>> Obviously, every additional license
>> inside Squeak creates a risk of 'contamination', and this is a
>> combinatorial explosion you want to avoid.
>
> Right. At least we want to avoid it in the core.
>
>> So, there are good reasons to try to keep at least the basics 
>> ('kernel',
>> 'coder', maybe even 'carnival') clean (SqueakL-code only).
>
> Right.
>
>> What I'm arguing here, and what caused this whole discussion it seems
>> ;-), is that there are alternative licenses that are strict supersets 
>> of
>> the SqueakL(*).
>
>> There is also public domain code - code, where the
>> original owner has relinguished all rights (including the copyright) 
>> to
>> the work and everyone is free to do as they please with the work. Code
>> under these licenses (I think even refactory.com's 'click here to
>> download the Refactoring Browser' constitutes a license - they keep 
>> the
>> copyright by default, so they must grant some rights based on this
>> privilege by offering copyrighted works for download; a safe 
>> assumption
>> is that the BSD/MIT terms apply: use as you like, don't sue us), being
>> supersets of the SqueakL in terms of granting you at least all the
>> rights of the SqueakL and not burdening you with more obligations, 
>> could
>> be considered for inclusion into Squeak.
>
> Yes, I agree - there is a "license" from refactory.com. (I just read up
> a bit more http://faqs.org/faqs/law/copyright/faq/part2/)
>
> This is the interesting question. So you mean that we should allow
> multiple licenses in "official Squeak core"?
>
> Wouldn't this create a big messy pile of licenses that possibly
> conflict? I mean, who am I/you to say that MIT/BSD doesn't conflict 
> with
> Squeak-L? I agree - it looks like they don't but hey... :-)
>
>> I'm not saying that, therefore, this code should be willy-nilly 
>> included
>> into Squeak. Of course, it is preferable to go to the author, explain 
>> (a
>> Swiki page might be in order) this whole thing here and discuss the
>> issue of dual licensing; however, *iff* the author refuses to do so 
>> and
>> *iff* the code is deemed important enough, I would vote in favor of
>> nevertheless including the stuff.
>
> And then nothing is stopping this scenario from happening. I am not 
> sure
> I like the potential outcome.
>
>> Short summary: please keep contributing, Stef :-)
>
> Definitely. There is no reason whatsoever to stop contributing - Squeak
> is AFACT on pretty sound ground licensewise. We just want to make it
> better and simpler! :-)
>
> regards, Göran
>