4 ms·
>and went after anyone who tried to make a clean-room implementation of it. That doesn't sound legal. You're allowed to clean-room implement anything aren't y
by loewenskind 16y ago
>and went after anyone who tried to make a clean-room implementation of it.
That doesn't sound legal. You're allowed to clean-room implement anything aren't you?
- avar 16y agoThe idea was that their network protocol was an original copyrighted work, and that any clean-room implementation of it would be a derived work. That does indeed sound crazy, but they also had piles of cash to throw at lawyers to go after anyone who challenged them, and (maybe as a result of that) nobody ever made e.g. a clean-room BSD or LGPL licensed version of the mysql library. So it never made it to court AFAIK.
- jrockway 16y agoA nice fiction. The reality is that copyright covers verbatim copying. If you read a book and write a review, the term "derived work" makes English sense, but not legal sense. Similarly, if you read the MySQL code looking for magic words to say to the server to get it to give you data, saying those words is not copyright infringement. It's "learning", not "copying". If anyone used MySQL, I would write a BSD-licensed API just to prove a point. (And to be non-blocking.) But the world has moved on since the Oracle deal...
- eru 16y ago> Similarly, if you read the MySQL code looking for magic words to say to the server to get it to give you data, saying those words is not copyright infringement. It's "learning", not "copying". What would happen, if the protocol would require sending a picture of a picture of Mickey Mouse with each request (or some copy-righted novel)?
- jrockway 16y agoYou publish a benchmark that says how inefficient this makes MySQL, and then that "feature" goes away.
- eru 16y agoI was more concerned about the copyright repercussions on clean-room reverse engineering the protocol. Obviously the protocol includes copyrightable information, now.
- jerf 16y agoTo the best of my knowledge, the best answer is "nobody knows". There are some vaguely similar things in the precedents, especially when it comes to protocols that require you to use a particular patent to talk to a given device, and IIRC there is some verbiage in some places that distinguishes copying around for simple transmission reasons rather than delivery to an end user (though I can't imagine how to stretch that to cover this case), but in this particular case, nobody knows. In fact if you spend much time thinking about this it isn't at all hard to come up with a lot of scenarios that law doesn't have an answer for, in theory or in practice.
- eru 16y agoThanks!
- fragmede 16y agoThat's been tested in court before, in Sega v. Accolade (http://en.wikipedia.org/wiki/Sega_v._Accolade http://en.wikipedia.org/wiki/Sega_v._Accolade). (You had to display a Sega logo before your game could run, so you had to have their logo in your code) In short, it's fair use.
- InnocentB 16y agoMy understanding from the Wikipedia article you linked is that you just had to have the string SEGA in a specific spot on your ROM, and that the logo was in fact part of the Genesis III BIOS. A four-character string and an image are pretty different, copyright-wise.
- piotrSikora 16y agoThat's incorrect, just check libdrizzle.
- spudlyo 16y ago... a clean room implemented, BSD licensed MySQL client library. I believe MySQL/Sun used to make a lot of money dual licensing the GPL client library. No longer!
- mahmud 16y agoNot only, but even reverse engineering of proprietary protocols from packets in transit is legal, IIUC.