Hacker News Reader

Comment by tachyonbeam | Hacker News Reader

It's bad for programmers everywhere. If Oracle wins, you couldn't reimplement a compatible version of anything that's not open source. You couldn't make a third-party implementation of a corporate-owned programming language without putting yourself at risk.

Imagine you're a company using some Oracle database, and you want to reimplement some component of that system in order to migrate and escape their vendor lock-in... You can't, not without taking a huge legal risk.

In the short term, it's a victory for huge litigious corporations like Oracle. The good thing, I hope, is that in the long term, this will make people distrust closed source software even more. What Oracle wants, and has always wanted, is to bind you in the worst possible form of vendor lock-in. Open source and open standards are the answer. Refuse anything else.

View on HN

> If Oracle wins, you couldn't reimplement a compatible version of anything that's not open source

It's worse than that. OpenJDK, including the java standard library, is GPL. It's open source.

Oracle is claiming that even though source code defining the API is open source, they retain copyright to the API itself.. regardless of the open source implementation of the API by themselves.

If this case goes through, you won't be able to mimic an open source project's API unless you also verify they provide the API under a similarly permissive license.

> Oracle is claiming that even though source code defining the API is open source, they retain copyright to the API itself..

Are you sure they are claiming that? A weaker claim could be that Google's usage does not conform to the GPL. Indeed if you were correct and the ruling is in Oracle's favor then the GPL is broken as designed, I haven't heard anyone express that concern in connection with this case.

> A weaker claim could be that Google's usage does not conform to the GPL.

At the time at issue, Java wasn't under the GPL. That Google did not have a (public offer or private) license to Java is not in dispute.

I very much doubt that anyone in big tech wants to open the can of worms that is GPL enforcement.

> OpenJDK, including the java standard library, is GPL. It's open source.

The code at issue is pre-OpenJDK, and wasn't open source when used (though much of it may now also be available in OpenJDK, not that Google uses it under the GPL, or claims to, anyhow.)

Sun released Java under the GPL in 2006 and most of JDK in mid-2007, but the first release of Android was in 2008. Or is the issue that they were developing it behind closed doors while it was still proprietary, even though they never distributed it to anyone?

The first public release of Android was a beta in 2007 (also, layer in 2007, the first SDK preview was released), and it was released to OHA members before that.

But the GPL is irrelevant, Google has never distributed under the GPL terms. That whatever it was doing was not under a Sun/Oracle license is not in dispute.

Yeah I agree the GPL is not relevant to the case at hand, though it does raise the question whether (assuming Oracle wins) the logical conclusion is that the Java APIs are under the GPL/Proprietary dual-license Oracle offers today or that the copyright license itself doesn't apply to the API and a separate license is needed -- in which case I guess it's high time for GPLv4.

Except anti-GPL folks have made GPL almost irrelevant outside Java, Linux kernel, GCC and a couple of GNU tools.

That's because the GPL is not as Hippie as you might think.

Your code is just protected if it gets re-distributed, that's why google never ever has to opensource their "Server OS/Kernel".

Also they changed GCC to GPLv3 which is just a super stupid decision for a Compiler.

by the by... wasn't Oracle's original product a relational database management system (RDBMS) that offered a SQL API... which was invented/copyrighted by IBM ?

https://en.wikipedia.org/wiki/SQL#History

Of course.

But Oracle doesn't care one bit about anything else that's not Oracle.

Though it is still to be known how many reimplemented APIs from others does Oracle have in their products

Nope, they pay for ISO which Google hasn't done for Java.

You don't have to pay for ISO to implement an ISO standard. In addition, Oracle released their database software in 1979 but SQL only became an ANSI standard in 1986 (and an ISO standard in 1987) so they did originally reimplement an API which wasn't a public standard.

Sure you do, because the ISO documents aren't available for free nor the certification process.

I'm pretty sure that a fair number of commonly used QR libraries were not developed by people who shelled out for a copy of ISO 18004 (they probably just looked at descriptions of how it works outside of the standard document). Similarly, you can implement your own SQL dialect without needing to read the ISO standard for it.

Yes if you want to be certified and want to get a copy of the standards document you'll need to pay, but it's not necessary (and I doubt that Oracle did, given that they implemented SQL before the standard existed and probably were involved in drafting it).

Sure that is how 1€ product shops work, which doesn't mean the final product actually does what it supposed to do.

Then that screwdriver turns itself instead of turning the screw.

> probably were involved in drafting it

Which also isn't done for free beer, ISO processes have associated costs.

I'm not disagreeing with the validity of ISO or anything like that, I don't know where you got that impression.

The original point being made in this thread was that Oracle copied the API of SQL from IBM and thus is guilty of the same thing that they are accusing Google of doing. The counter-point was that SQL is an ISO standard and Oracle (allegedly) licensed it, which was then followed up by my point about Oracle DB predating the ISO standard. Hence why ISO is being discussed and why I was making the point that buying a standard document from ISO or paying for certification are not really relevant to this discussion.

Some bmc commentary on that:

https://www.youtube.com/watch?v=-zRN7XLCRhc&t=2280s

It's the whole "Disney wants the copyright laws to prevent anyone from doing to Disney what Disney did to the Brothers Grimm" thing.

Yes and no. Yes, Oracle was a database with a SQL API. Yes, SQL was invented by IBM. No, it wasn't copyrighted by IBM, because at the time, nobody thought you could do that.

I thought you couldn't 'copyright' something?

I understood copyright to be something you had if you had created it.

Ah, brain cramp. I was thinking of patent, not copyright. Until State Street, nobody thought you could patent software, so nobody tried.

There was still controversy about software copyrightability in 1974. That said, I expect IBM still explicitly copyrighted their implementation of SQL (and licensed it [1]) as they had first started doing with some of the System/360 software during the prior decade. Note that, in 1974, you still had to explicitly copyright something; it didn't just happen automatically like today.

[1] As a side note, IBM may not have been the first to license software but they certainly helped make it a standard practice. When they unbundled much of the System/360 software, they felt that copyright was a weak protection (given its uncertain legal status) so they combined it with a license.

> it didn't just happen automatically like today.

... in the US of A.

Berne convention started a little bit over a full century before that.

Impressive that the superfluous “need” for the Copyright (c) YYYY notice still persists to this day in the mind of many.

There can still be advantages to being explicit as I understand it to enforce copyright/collect damages. For example, most people don't actually deposit copies in circumstances where they're technically supposed to.

Pre-1976, authors had to explicitly add a copyright notice. The Copyright Act of 1976 changed all that. Creative works automatically gained copyright the moment it was created.

SQL is an ISO standard, which Oracle like everyone else also pays for.

You can make an open-source database that uses the SQL language without paying a dime to anyone or signing any contracts. You don't have to pay for the ISO standard document to implement it - you can just infer behavior based on what you already know about sql.

The state of SQL databases would be better if more people did pay for and try to actually implement the standards as written... same goes for many of these expensive standard documents.

I’ve made a life long habit of collecting and archiving these in my personal documents whenever I get my hands on one. Because these are valuable guidelines as to how things are supposed to work.

Which isn't what ISO says it should be.

>you want to reimplement some component of that system

That's patching and is already disallowed by licenses in USA, but allowed by law, but that law was goofed up by a court.

Could you be more specific? What law allows it, and how was it goofed up by a court?

It's allowed by DMCA, but a court ruled that the license overrides law, and now all licenses have "don't patch" clause.

Let us assume for a moment Oracle wins this so what? Android will become illegal and then Oracle would ask Google to do what? Pay royalties? Pay Damages? Ask all the android programmers in the world to stop using Android ? Abandon android runtime? I think what is gonna happen realistically is capitalism will take over they will settle it among themselves and share user data as a bargain probably get some royalties. The worst that can happen is Google will abandon android runtime or transfer the development to Oracle. Either way nothing much is gonna happen. Android has already won now say from tomorrow Oracle develops and owns the android development toolchain, so what, from a programmers perspective nothing much.

Nothing much? This isn't just about the Java API. This is about EVERY API. You might not be able to write a web server anymore, depending on who owns the copyright for HTTP. You didn't invent GET, PUT, and POST, you thieving copyright abuser.

Back when I was working for Nokia, the APIs were patented, this is nothing new.

Patents and copyrights aren't the same thing.

Patenting an API makes sense, since APIs are essentially ideas. Copyrighting an API does not make sense.

You can't patent an idea. You can patent the implementation of it if you're specific enough.

Well sort of. You don't need to have built the actual thing to get a patent.

You cannot patent the completely generic idea of "a tool to lift people in buildings", but you can patent an elevator.

Similarly, you cannot patent "the ability to scan a database", but you can patent an api for doing so, even without a complete implementation of that api.

> You cannot patent the completely generic idea of "a tool to lift people in buildings",

You should see some of the software patents that get approved nowadays to see just how often "idea on a computer" gets greenlit.

Like that Sony patent from 2005 to control the brain without conducting any experiments to confirm if that even works:

https://arstechnica.com/uncategorized/2005/04/4785-2/

You and your Stockholm syndrome.

Nope, I just don't live from rainbows and happy music tunes.

Quite interesting claim as Nokia was still an EU company back then and software "as such" can't be patented to this day in the EU. (That's sadly not true in practice any more for some time by now as more and more software patents are granted under the notion beeing "computer aided inventions". This also overlaps with strong political will to finally lift the general ban on software patents. Thankfully some people are fighting this nonsense though[1]).

[1] https://fsfe.org/activities/swpat/swpat.en.html

I dont think so, Webserver are based on HTTP protocol which is an Open web standard published by IETF. https://tools.ietf.org/html/rfc7231 It could be copyrighted but its usage doesn't mean you are thieving it. For instance "Linux" is a copyright owned by Linus Torvalds, Redhat isnt thieving it. Your understanding is incorrect here.

> It could be copyrighted but its usage doesn't mean you are thieving it.

Says who? Oracle is claiming exactly the opposite, and if they win this will become "law" in the USA. I think you've missed the whole point of this issue here.

> For instance "Linux" is a copyright owned by Linus Torvalds…

For starters: Linux™ is a trademark, not "a copyright".

> Your understanding is incorrect here.

It seems, at least to me, that you actually missed a few core point of the whole story. ;-)

Reply on news.ycombinator.com