svn commit: r421549 - in head: . Mk

[Kubilay Kocak]

On 9/09/2016 6:35 PM, Alexey Dokuchaev wrote:
> On Fri, Sep 09, 2016 at 06:29:33PM +1000, Kubilay Kocak wrote:
>> On 9/09/2016 4:26 PM, Baptiste Daroussin wrote:
>>> In both case that means there is NO license and then we should
>>> not distribute them at all.
>> 
>> There are cases where software has no license, the author says so,
>> but they mean, and/or say 'free to do with what you please'. This
>> is neither NONE, nor undefined (in terms of the 'terms'), nor PD,
>> nor 'empty(LICENSE)'.
> 
> That's why I prefer something along
> UNCLEAR/MOOT/VAGUE/CONTROVERSIAL/etc. to cover all those "weird"
> cases and be done with it.
> 
> ./danfe
> 

And precisely why UNDEFINED was suggested over NONE.

The reason for UNDEFINED over others? Not as prescriptive or subjective.
More inclusive (better coverage/utility).

What's really important is obtaining/knowing the 'terms' (LICENSE_PERMS)
of (re)distribution, not knowing the license name.

LICENSE itself doesn't technically need to be defined or definable for
the software to 'have terms'.

In fact, for well known licenses with standard terms, defining the
LICENSE in a port is enough to state the terms. It allows us (porters)
to skip defining PERMS. That's really the variables only utility.

If the software author says 'no license, go do whatever you want', you
can do:

  LICENSE=UNDEFINED
  LICENSE_PERMS=<whatever>

Whereas:

  LICENSE=NONE
  LICENSE_PERMS=<whatever>

Doesn't really grok.

Alternatively, for the 'don't know terms at all' case:

  LICENSE=UNDEFINED

No LICENSE_PERMS set, we know what to do: limit (re)distribution