BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 24, 2024 23:13 UTC (Fri) by khim (subscriber, #9252)
In reply to: BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog) by mb
Parent article: BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Kernel is GPLv2, but metadata is not.

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 25, 2024 12:25 UTC (Sat) by tuna (guest, #44480) [Link] (5 responses)

I would argue that the metadata is either derived work from the GPL2 licensed data or mere aggregation that is not worthy of copyright in most legal jurisdictions.

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 25, 2024 12:35 UTC (Sat) by khim (subscriber, #9252) [Link] (4 responses)

You argue… on what basis, precisely? Copyright in compilation exists precisely to ensure that copyright for metainformation exists separately from copyright for kernel source. And “mere aggregation” implies that it's something that can be easily collected by someone else.

Also, you have the exact same situation that is happening today: RHEL kernel sources. Go to the court, prove that you have the right to have the because they are “derived works” of kernel, make IBM publish them… then you would have a case.

I'll wish you luck because that would, obviously, never happen.

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 25, 2024 17:56 UTC (Sat) by tuna (guest, #44480) [Link] (2 responses)

The RHEL kernel (Linux) packages are definitely licensed under GPLv2. IBM does not have to publish them on the internet anywhere, but they have to provide the source code to they people they have provided the binary packages (if they ask for them).

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 25, 2024 18:08 UTC (Sat) by mjg59 (subscriber, #23239) [Link] (1 responses)

This is a common misinterpretation of GPLv2 section 3. Under 3(a), you must provide the source code alongside the binary code. In this scenario you only need to give the source code to people who get the binaries, but you have to do it simultaneously (general interpretation is that it's fine if the recipient chooses to take the binary but not the source, so eg having the binary and source packages on the same download service is ok - you don't need to include the source in the binary download). The alternative is 3(b) where you only have to provide the binaries as long as you provide the source on request - but the trick here is that it's open to *everyone*, not just people you've given the binaries to. IBM may be distributing under 3(a), but in that case they can't merely provide the source on request, it has to be available alongside the binaries.

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 26, 2024 8:48 UTC (Sun) by tuna (guest, #44480) [Link]

Thanks for the clarification!

BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Posted May 25, 2024 23:30 UTC (Sat) by jjs (guest, #10315) [Link]

I'm not certain copyright of compilations protects metadata - from reading the Feist decision, it appears to cover the copyright on the entire product - i.e. the structure, composition, and selection.

It appears metadata, assuming it's original and creative works sufficient for copyright, would be protected by their own copyright by the writer/developer.

IANAL, my reading of the decisions.