The tech industry runs on myth and superstition more than most people will admit. One persistent myth is the special role of "linking" in propagating "derived work" copyright status. Are we really supposed to believe that if non-copyleft code A is tightly coupled with copyleft code B, then the copyleft
does propagate to A if we summon the ld-linux.so.2 demon for dynamic linking and
doesn't propagate if A uses B via system(2) and a pipe? Ridiculous. Whether A is a derivative work of B has to do with the extent to which A is coupled to B in particular, not the mechanism of this coupling.
Personally, I've always found it hard to believe that merely using, e.g. libreadline (or Linux EXPORT_SYMBOL_GPL) really makes one program a derivative work of another. Is there any actual legal precedent for mere dynamic linking propagating copyleft? Common sense suggests that, no, linking to libreadline or libmysql or whatever doesn't make a program that does something totally different a derivative work.
Has the linking-propagates-copyleft theory ever actually been tested? Might we have been making life hard for ourselves for decades for no actual reason?
Personally, I've always found it hard to believe that merely using, e.g. libreadline (or Linux EXPORT_SYMBOL_GPL) really makes one program a derivative work of another. Is there any actual legal precedent for mere dynamic linking propagating copyleft? Common sense suggests that, no, linking to libreadline or libmysql or whatever doesn't make a program that does something totally different a derivative work.
Has the linking-propagates-copyleft theory ever actually been tested? Might we have been making life hard for ourselves for decades for no actual reason?