There's probably a better case for compilers being fair use than this (not that this doesn't have a great case).

I'd be more concerned about emulators and actual chips. People have been assuming that copyright doesn't apply to APIs for so long that I'm sure there's any number of cases of accidental "copyright infringement as defined by Oracle". I'm just not sure that compilers are a prime example (at least as relating to instruction sets).

"Probably". Hah. That just shows the ramifications of Oracle winning this case are fucking insane. We've been operating as if this stuff isn't copyrightable for the last 50 years. You may as well burn the whole industry to the ground if they have their way.

> You may as well burn the whole industry to the ground if they have their way.

At least in the United States. China, on the other hand, will laugh at us as they lay claim to everything American lawyers prevent us from doing.

In EU interoperability is a valid reason to break copyright in this manner.

It would matter here too though as we do also want to sell software to US.

But if one stays out of US markets then things are as they have always been.

The EU has similar cases occur. For example between Lego and Best-Lock over what features of lego are copyrightable versus which are functional (and their competitors' minifigures are also designed to "interoperate" with lego bricks)

https://thetmca.com/equitable-estoppel-defense-denies-lego-f...

Yeah. As can be seen in that case they are focusing on things that are mandatory (non copyrightable) and those that are just stylistic choices (copyrightable).

API compatibility doesn’t have any stylistic choices in it.

Could you de-copyright things in this manner? Say I make locks that only open when they hear music from Disney's lion king. Are you allowed to play that song to open the lock, even without opening it? Does it matter if it was disney or me that made the lock?

Silly example of course, can't think of any better. Feel free to substitute your own.

For anyone to care you'd have to be an adversary of the manufacturer who published the specification. Generally they publish these details in order to encourage people to port OSs and compilers.

If a tree infringes in the woods and the silicon manufacturer doesn't hear it...

Then you, for example, write a compiler for SPARC in 1990 and distribute it for twenty years, only to see Oracle buy Sun and then decide they'd rather sue you for twenty years of copyright infringement than maintain your cooperation in continuing to produce a compiler for a processor architecture they're planning to discontinue.

I think emulators and competing CPUs are a more direct analogue.