For Approval: Transitive Grace Period Public Licence, v1.0

Tzeng, Nigel H. Nigel.Tzeng at
Fri Feb 20 14:50:25 UTC 2009

From: Russ Nelson [nelson at]

>Tzeng, Nigel H. writes:
>>>And it would be a damn shame for the OSI to approve a license as 
>>>open source that had that property.
>> Yes.  It would also be a damn shame if the OSI never bothered to 
>> approve a license that was "different" and perhaps even "complex" 
>> even if it appears to adhere to the spirit of open source and provided 
>> something new.

>The problem is that the license either 1) fails to have the requested
>transitive nature (in which case a waiver suffices), or 2) turns
>OSI-approved open source licenses such as the OSL into time-restricted
>license.  It would totally and completely break the idea of OSI
>approval on a license.

>So, yeah, "appears to" is the operative phrase.

Lets take your example:

>Not.  At.  All.  Let's say that I get a copy of some TGPPL code.  I
>immediately release it under the OSL.  What happens to that code
>release in a year?  I can think of two possibilities:

>1) The OSL remains in force, in which case, the TGPPL is really the
>GPPL; its transitive nature being completely lost.  Note that I could
>also do this on day 364 of my 12 months.
>2) The OSL expires, except that the OSL says it doesn't expire as
>long as the copyright remains in force.  That means that the code was
>never *really* licensed under the OSL, at which point the judge would
>throw up his hands and say "You're all completely insane; get out of
>my courtroom."

Neither possibility. 

If you are not the copyright holder you can't release it under the OSL 
(or BSD). Either you have source previously released under TGPPL 
or you have someone's currently proprietary unreleased source.  

If you are the copyright holder then you've just dual licensed your
portion of the codebase under OSL (or BSD) and TGPPL.  The other 
previously released code is still under TGPPL.  That you must release 
under TGPPL within 12 months is a given because it's a requirement 
for the license on the code you reused.  

Transitivity is preserved although you've just opened the window for 
forking your contributions.

What you license your code for the 12 months is (mostly) immaterial.  
It's your code. What happens after is what happens after any other time 
you have two licenses on the same code.

Any other degenerate case I can think of applies to the attempted mixing 
of any two incompatible open source licenses.

>> Essentially, it appears folks may have accepted that there is no middle 
>> ground between BSD and GPL.

>Mmmmmm, no, it is that Zooko's middle ground does not work.  And
>remember: Zooko insists on transitivity.  That is not necessarily part
>of a middle ground.

Actually, what it appears is that folks are creating a series of scenarios
that appear to fail and declaring failure without providing him time to
show how it might work.

Your other two objections:

Someone takes TGPPL code, makes a product and never releases
the result.  You don't know so you cant sue.  But this is no different 
than non-compliance for GPL or any other open source license.  
There's no expectation for a thief to inform the upstream "Hey, you 
need to sue us!"

The other objection was that you only get a trickle of code that is 
a year old.  This is incorrect.  If they made a binary release a year 
ago, all the code for that release must now be made available.  If 
they want, they can trickle out development code a year old but 
that's no different than any other project.  They aren't required to 
release development code because they haven't distributed until 
they've released binary.

I've seen open source code bases that are completely unbuildable 
from their repositories.   Mostly it's dependencies on old open 
source libraries that you can't get a hold of anymore but they 
still have the binaries on their dev machines.  They can't distribute 
those (they didn't keep the source) and the version 0.2a they 
used is no longer anywhere and version 1.4 is structurally 
completely different.

I'm not the submitter and I can honestly say I was a skeptic at first but
the examples provided as to why it doesn't work so far has made some
interesting assumptions that I believe to be inaccurate.


More information about the License-review mailing list