Re: Three problems with Chinput
On Fri, 2 May 2003, James Su wrote:
> Then, how do you think about the relationship between the linux kernel
> and close sourced modules, like NVidia kernel driver? Don't they obay
> those two rules?
Ha! You don't want to use binary-only kernel module as an example because
it is a beast of its own. This is a very gray area and subject to heated
debate many times a year. AFAIK, there's no legal consensus about this
issue. It's just that Linus have gave his personal blessing to the binary
only modules. However as Alan Cox pointed out in an email sent to lkml
about 2 years ago that "Linus opinion on this is irrelevant". He also said
in the same email that "Anyone releasing binary only modules does so
having made their own appropriate risk assessment and having talked (I
hope) to their insurers". In the 2.5 series a EXPORT_SYMBOL_GPL is
introduced such that a new internal kernel interface can be marked as only
available to modules with a GPL compatible license should the developer
decide to do so. Practically this will make binary only modules even
harder. So the future of binary only modules is uncertain.
>From what I understand the basic assumption about a binary only modules is
that it can not use any header file, data structure, macros, etc. from the
kernel source. If you can do that, the module itself is not a violation of
anything, albeit non-free. You can redistribute it as long as the author
permits you. Only when you link this module into kernel, the kernel
becomes non-distributable! So no one does that.
The crucial difference between kernel module example and your case is that
your pinyin module obviously uses data structures and macros from scim
proper which is covered by LGPL. So even the module alone is not legally
distributable.
Regards,
rigel
Reply to: