On 14-10-22 06:16 AM, Kern Sibbald wrote:
[...]
Yes, copyrighting APIs would make interoperability impossible or
almost impossible.
In the appeals court decision, one of the reasons they ruled against
Google was that Android (Dalvik) wasn't compatible with Java. I realize
that many on this list probably would say that it's compatible, or
compatible to a degree, but the appeals court found differently. Did
Alsup state whether or not Dalvik was compatible with Java? There's
also been an assertion on this list that Dalvik contains reverse
engineered code; plenty wasn't reverse engineered. Apache Harmony:
reverse engineered; Dalvik: not.
The appeals court goes out of its way to document how and why Android is
incompatible, and explains exactly how Google could've avoided this
whole mess: reverse engineering.
In short, if the purpose is interoperability, it's absolutely, I think
almost explicitly stated by the appeals court, that it's fine to reverse
engineer. I believe it's an actual law that it's permissible, no
precedent required, not subject to what a judge has to say about it.
There are no interoperability concerns, provided reverse engineering is
employed. Provided.
I realize that Google has given the impression that they reverse
engineered, but no where in their appeal to the Supreme Court do they
actually make that claim -- the omission is notable. The silence speaks
for itself. Instead, Google dances around the topic, brings up reverse
engineering, says it important, and that's fine. I agree with their
appeal to the Supreme Court, it's all fine sentiment and sounds good.
If Google were go ahead and claim that they reverse engineered, I would
say "go for it," but they'll never make that claim. Their appeal
doesn't connect what they did, copying, to what's permissable, reverse
engineering.
It would be interesting if one of the Supreme Court justices were to ask
Google: did you reverse engineer, yes or no?
I agree that most people believe that APIs are not copyrightable and
we in open source have always assumed this is the case. However,
certain APIs are copyrighted and they are generally protected by not
disclosing the API -- such as Oracle's DB backup interface. Bacula
Systems (open core) was able to write an Oracle DB plugin only because
we signed an NDA.
I believe that most APIs are a unique creative work, but at the same
time, I believe that they should not be copyrightable otherwise, it
permits monopolies, and would effectively lock out open source.
In the appeals court decision, those judges go out of their way to
explain how Google could've avoided this whole hassle had they just
reverse-engineered. Again, Google didn't reverse engineer, it's as
explicit as possible in the appeals court decision.
If the concern is about using an API, that's covered by fair use.
Google's lawyers and PR has very cleverly implied that "oh no, you can't
use a copyrighted API," which is a misnomer. Nothing prevents using an
API exactly as up until now, what would be different is that copying the
SSO of an API wouldn't be allowed.
[...]
Bottom line: as an open source author and advocate, I would not like
to see published APIs copyrightable.
I'm not so sure that I'm in favor of making the SSO of an API subject to
copyright protection, but that's perhaps a red-herring, or maybe a
straw-man. What are the implications, either way, for the GPL?
Not Alsup, not the appeals court, have suggested copyright on an actual
API. If the SSO of an API is original and creative, and meets the
other criteria, then that the SSO of that API might be protected by
copyright, according to the appeals court logic.
Again, even if the SSO of an API were protected by copyright, fair use
still allows the usage of the API itself, and reverse engineering is
still permitted, exactly as before. No change at all to either of
those; the sky isn't falling if the appeals court decision stands.
Alternately, giving the greenlight to copy the SSO of an API, which
means, literally, verbatim copying code (declaring code), and farming
out the implementation, that would be the tipping point because it's
sufficiently cheaper and faster than reverse engineering; and, for all
practical purposes, only really applies to GPL protected software. ASL
software, realistically, won't suffer from this.
Google copied the declaring code -- how can that possibly be ok? They
didn't reverse engineer.
-Thufir
