Henning Makholm wrote:
It seems that my bad English got me: I used the word "basically" in the same sense that its cognate can be used in Portuguese, "in the basic/normal cases...". I apologize for the confusion.If it is wrong to begin with, then it is also basically wrong.
Which I now solemnly amend to "The kind of bits you normally (>99% of the times) find in .h files in c-language based projects, and often (>50% of the times) find in .h files in c++ based projects, are those defining interfaces, deeming them uncopyrightable by current USofAn and Brazilian law". Better?Well, yes. And which conclusion do you draw from that observation?
It is my most humble, but firm opinion:1. That a plugin (written in C or C++) that includes an GPL'd plugin-interfaces.h file, and uses the interfaces described there for its normal chores, is not a derivative work of the GPL'd "I can load plugins" program.
2. That said plugin, when in its compiled form, if it contains at all any inlined functions or macros from the GPL'd plugin-interfaces.h file, is merely a volume of storage or distribution in accordance to the disposition on the GPL, section 2, paragraph 3.
2.a. That is, as long as it does not mess with implementation details of the plugin-loading mechanism, attaining to the use of published interfaces -- if it doesn't, pieces of the implementation will "leak" past the Filtration phase and it can be deemed a derivative work on the plugin loader.
3. That the plugin's licensing is completely independent of the "I can load plugins" program's one, and that it can be licensed as the author sees fit. Which, of course, includes the QPL.
4. That, reinforcing, no dispostion in the GPL would stop the user from linking dynamically at run-time said plugin, to make possible its use.
HTH, Massa