1. 31

    1. 5

      Ahh thanks. @jcs maybe that link should replace this one.

      1. 5

        It is so.

      2. 1

        What about a link with some line breaks? :-)

      3. 3

        Um. Does it mean Unix-like systems are illegal now? It’s all Bell Labs' APIs, after all.

        1. 7

          No - a lot of that was decided as part of previous cases; https://en.wikipedia.org/wiki/UNIX_System_Laboratories,Inc.v.Berkeley_Software_Design,Inc. was the original lawsuit in 1992, which settled that BSDs are not derived from Bell Labs Unix. Despite several other lawsuits, not much has materially changed since then, and https://en.wikipedia.org/wiki/SCO%E2%80%93Linux_controversies#UNIX_copyrights_ownership has a good summary of where things have stood since 2008.

          The particularly upsetting thing about this latest holding is that it affirms the novel element of claiming that copyright applies to symbol names (if one believes the rumors on Twitter, it applies even to type signatures with names omitted, but no technically-savvy lawyers seem to have written anything about that just yet). What that means is that any independently-written code which links against somebody else’s copyrighted code through its published API is, apparently, suddenly a derivative work. This is regardless of static or dynamic, since it’s the names that are the infringing element. This also applies to clean-room re-implementations of an API.

          This assertion was not made in the earlier cases, and it’s dubious that they could be re-tried in light of the changing law, but I’m not a lawyer.

          It’s extremely doubtful that the court had any understanding that if anyone takes its ruling seriously, it is now impossible to write software that interoperates with other software until a licensing framework that has never before existed is put into place.

          I should probably refrain from speculating on what happens next.

          Edit: The Lobsters Markdown code apparently doesn’t include a trailing period as part of a link; this one needed it. It’s now in an explicit […](…) construct.

          1. 4

            According to the EFF, “Today’s decision doesn’t mean that Oracle has won the lawsuit. The case will now return to the district court for a trial on Google’s fair use defense”. Fingers crossed for a ruling in favor of fair use.


            1. 1

              Yeah. Which is good, certainly, but the holding that was affirmed is almost certainly more important than who pays penalties to whom, and stands even if the parties settle. That said, the court that issued it doesn’t necessarily set precedents for other circuits, and this case being appealed to it was only possible due to some unique history.

            2. 3

              Those lawsuits were about source code though, which is not what I understand as APIs. I was under the impression that creating compatible software (thus, compatible API reimplementations you mention) was specifically legal, and Unix-alikes are full of specifically that (though it’s a POSIX standard, which may or may not matter, IANAL).

              Also, what about GNU and other java implementations? Are they any different from Android in this respect?

              1. 5

                This time around, the case was specifically about that distinction, since the alleged infringement was, in fact, a clean reimplementation. I’m sure if there were any existing case law about that, it would have come up in the filings - and perhaps it did; I have yet to see commentary which goes into any real depth, and the filings are hundreds of pages long, and excuses excuses. :)

                So I can’t answer as to whether reimplementation was explicitly legal until now, vs. legal by default because nobody argued against it before. The obvious software I can think of that it would have been tested on is Wine, which of course reimplemented the Windows APIs at a time when there was very little awareness of what open-source was at all. But from a quick search, it doesn’t appear they’ve ever been sued (difficult to tell for sure, since there have been a lot of lawsuits about, you know, the alcoholic beverage). At the risk of being speculative, I believe Microsoft would have sued Wine if they thought they could prevail. I know that the project has always been extremely careful about not using information that one has to sign an agreement with Microsoft to get, and I’m sure that helps.

                1. 3

                  nobody argued against it before

                  I think there was something about file formats (MS Word?) and/or protocols (ICQ?) being legally reimplementable, because competition free capitalism market king. But not APIs? That’s peculiar. But, as a non-USAian non-lawyer, I don’t know what I’m talking about.

                  Wine, which of course reimplemented the Windows APIs at a time when there was very little awareness of what open-source was at all

                  Initial release 1993? Please don’t say “open-source” about free software projects that started five years before the term existed :)

                  Around that time Stallman was advocating for boycotting Apple, who previously sued Microsoft for daring to have a window system, a concept Apple stole from Xerox PARC. But X and OS/2 were alright, for some reason. That, however, was about patents, not copyrights, so it probably doesn’t apply.

                  1. 3

                    The area around protocols seems to have gotten murkier, too. In the ‘90s, AOL was not happy about gaim, but they didn’t think they had a case regarding the protocol itself, so they limited themselves to trademark complaints about the name (one reason it was later renamed to pidgin), and to mucking with the protocol periodically to break third-party clients. But fast-forward to today, and you have companies like Snapchat and WhatsApp aggressively using the DMCA to take down third-party clients. Probably AOL would’ve done something similar today.

                    1. 2

                      Sorry. Point taken! And yes, I remember the GNU-Apple boycott. It was the reason I couldn’t start to learn about GNU software for roughly the next eight years. :)

                  1. 1

                    Your paste has the same problem - the trailing period isn’t included as part of the link. I fixed it above with […](…), but that’s not ideal. Obviously, matching URLs in text is a hard problem, and I don’t see a quick fix to the parser in this case.

                    1. 2

                      You forgot to add the period . to the (…) URL section of your link; you only added it to the displayed text in […]. Your link is still broken. Here’s a working one.

                      In general, you can avoid duplicating the URL by using <…> instead of […](…):


                      This also prevents the _ underscores in the displayed text from italicizing the text between them, which happened in your link.

                      However, when rendered on Lobsters, the above <…> syntax does not handle the ending period correctly. I have reported this bug in Lobsters.

                      1. 1

                        If you include trailing full stops you’ll get bad links in other cases. Honestly I think people should expect to have to use explicit markdown in cases where the URL is funny like that. Arguably Wikipedia should avoid having their URLs end in full stops - even though they’re technically allowed, they’re problematic for users.

                        1. 1

                          A nice example of how technical issues are usually also social issues. :)

                          Anyway, I don’t have a strong opinion as long as there’s a way I can make it work. I wish the failure had been immediately obvious, though.

                        2. 1


                          1. 1

                            Indeed :(

                  2. 2

                    This should certainly have the “law” tag.

                    1. 1

                      Argh, I knew I missed something