Hacker News Reader

Comment by rhino369 | Hacker News Reader

While I'm not sure how I stand on Oracle v. Google, I don't really agree with EFF's legal analysis here.

Nothing in that case suggests that using an API is infringement. It's about re-implementing an API for a different purpose. So using Twitters API is fine, but copying Twitters API for use on a Twitter clone service, not fine.

I also don't think this case necessarily prevents re-using an API for interoperability. The case was about Google reimplementing Java API for use in the android davlik engine. That really isn't interoperability since Java wouldn't run on it. They just copied some java apis for the convenience of android programers.

Pardon my ignorance but is that really common in the programming world? To take someones API and use it for a totally different purpose?

View on HN

It happens all the time, and that's a good thing.

> Twitters API for use on a Twitter clone service, not fine.

That's precisely what's bothering me the most. This is completely contingent on your definition of what is an API.

The judgement is setting a precedent in defining an API by its routes and not by what or how it processes data. To the court, it does not matter if you are doing data intensive computations or just stripping html from a string... if the route is the same you are screwed.

HTTP/1.1 GET /v1/news/latest_updates becomes intellectual property of Company Inc.

Imagine the amount of patent trolling wars that are going to be waged because of that ruling, what a freaking nightmare.

PS: The parent comment should not be downvoted.

- I come to HN to hear people's opinions and learn from them, even if they are dissenting. It's fine to disagree, but do not silence people for thinking differently. Every constructive contribution should be part of the discussion.

Again, pardon my ignorance, my only attempt at API use resulted in a very poorly programmed python app that was supposed to tell me which train to take . . .

but isn't that just using the HTTP/1.1 GET method? Just using someone else's API with a new paramter isn't an expressive action and clearly isn't copyrightable.

But the court would say the HTTP/1.1 GET [url] api is copyrighted by whoever wrote the HTTP spec.

The court didn't rule than ANY api is copyrightable. Just that the specific Java ones at issue where.

For example, Copyright law has what is called the "merger doctrine" which essentially says that if there is only one or maybe a few ways to implement something, then it isn't copyrightable.

So while you can copyright Microsoft Excel, you probably can't copyright a simple method that adds two intergers together. There is essentially only one way to do it.

The Fed Cir. in Oracle rejected that argument because there are unlimited ways to implement the Java libraries API. Why did Google even call them java.XXXXXXX.XXXX when they could have been Android.XXX.XXX, etc.

There is another doctrine that states that industry convention can't be copyrighted. GET is industry convention and would never be copyrightable. But the data structure it send back may or may not be.

> " Why did Google even call them java.XXXXXXX.XXXX when they could have been Android.XXX.XXX, etc."

That wouldn't be implementing the Java APIs any more, as evidenced by the fact that no code relying on the Java APIs would be able to compile against your hypothetical Android standard library. By the Java language's own definition, changing the package and class names makes it a different API.

> - I come to HN to hear people's opinions and learn from them, even if they are dissenting. It's fine to disagree, but do not silence people for thinking differently. Every constructive contribution should be part of the discussion.

While I agree with you, pg (in)famously stated that it was OK to down vote to express disagreement (https://news.ycombinator.com/item?id=117171). Would be nice to have dang change the "official" stance on this though.

>The judgement is setting a precedent in defining an API by its routes and not by what or how it processes data.

More generally, the ruling is bad because patents are supposed to be for an implementation, not an interface. For how you accomplished something, not the accomplishment. For a combustion engine, not "being able to get into a box and exiting at your destination in a short time".

Every outrageous parent grant seems to have this attribute: the problem with one-click is how the patent covered "everything that accomplishes one-click", not "our particular solution for getting one-click to happen".

Here, they're patenting the accomplishment of "making this call get this result", not "how we managed to get that response to that call".

The trial at this point is entirely about copyright, not patents

That really isn't interoperability since Java wouldn't run on it.

A Java package I wrote in 1996 is used in Android apps. So they had to recompile it or something, I don't know.

It happens. Just off the top of my head, recently Mozilla's Servo decided to copy Chrome's embedding API CEF.

using calls in an api for a different purpose is not copying by legal def of copyright law

I didn't mean using, I meant copying the API itself to re-implement in a different piece of software.

Reply on news.ycombinator.com