🚀 go-pugleaf

In article <170q0iINN84v@agate.berkeley.edu> jbuck@forney.berkeley.edu (Joe Buck) writes:
>
> I think that the GLPL is the appropriate protection to use for object and
> class libraries; this is consistent with what FSF has already done and
> will avoid confusion by having the same conditions on all Gnu-ish class
> libraries.
>
>   ...   For example, if you use libg++ in a
> project, you must distribute source to libg++.  You need not distribute
> source to your classes that are inherited from C++ classes (I purposely
> didn't say "derived" to avoid confusion with the concept of a derivative
> work in copyright law -- it isn't necessarily the same).  If, however, you
> make modifications to the source code of any part of libg++, you must
> distribute source to those modifications.
>

Ok, what about this scenario?

Let GWO be a gee-whiz object covered by GLPL

Suppose that I've obtained an object library containing Has-A
relationships with GWO.  Suppose again that there is a fundamental
algorithmic revelation resulting in a new and improved GWO-prime
which satisfies our previous interface specifications, is a order of
magnitude faster and twice as big.

Will:	our linkers do the right thing,
	developers be limited to Has-A-Pointer-To relationships,
   or
	I have to live with the old non-improved object and the
	  related object name-space pollution.


I think that the Is-A relationships will be worse, since they often
assume knowledge of the inner workings of GWO.


dwight shih					dwight@s1.gov