Sylvain Wallez wrote:
IANAL neither, but the ASF has a pretty strict internal practice to *not* to distribute any GPL code whatsoever in whatever form to explicitly avoid legal issues.John Morrison wrote:Sylvain Wallez wrote:IANAL, but the mock classes are created without referring to the real classes. They just contain the declarations of methods actually used by our code, and are made abstract as far as possible.
So I propose that we _remove all the conditional compilation stuff_ and write mock classes where needed to ensure that _all_ source files are always compiled.
Thoughts ?
+1 in general, my only missgivings are legal - where do we stand
creating 'mock' objects of somebody elses code that (I don't know
if this is a possibility) might be GPL (or other restricted licensing
agreement)?
Only thoses classes that are instanciated by our code need to be concrete class, and the methods body is then just a "throw new UnsupportedOperationException("this is a mock object);"
So it can be considered as a clean room reimplementation of a subset of the class public interface.
Any lawyer out there ?
In fact, nobody can even agree on whether calling a GPL java library using reflections and keeping method names in a configuration file can be called 'binding' or not. And this is also true for an LGPL library.
I'm sure the FSF would say yes and the library user would say no. But when RMS comes after you, that's where the fun stops.
In short, moch objects don't change the picture: the ASF doesn't distribute *GPL java code nor connects to it. (luckily for us, there is not much GPL java code out there... and it's easy to convince them to dual-license if we really want to use it because the ASF is a much bigger than than the FSF on java world)
--
Stefano Mazzocchi <[EMAIL PROTECTED]>
--------------------------------------------------------------------
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
