Weekly Edition
Return to the Announcements page
| |||||||||||||
Monty appeals Oracle's Sun merger (The Register)
The Register reports
that Michael 'Monty' Widenius has submitted an appeal against the European
competition watchdog's decision to clear Oracle's takeover of Sun
Microsystems earlier this year. "Michael 'Monty' Widenius, who sold his chunk of the MySQL database to Sun Microsystems in 2008 for $1bn, told various news outlets including The Register on Friday that he would make a full statement about the appeal once Oracle had responded to it."
(Log in to post comments)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 5, 2010 23:02 UTC (Mon) by stumbles (guest, #8796) [Link] Oh Monty, just give it up. You sold out making a wad or two. So now you are trying to have your $1bn AND eat your cake. You gave up that right by accepting your bag of coins.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 5, 2010 23:19 UTC (Mon) by stewart (subscriber, #50665) [Link]
Just to be clear, the $1billion was the total sale price, not what Monty got.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 4:12 UTC (Tue) by jordanb (guest, #45668) [Link]
I bet you could make some mighty fine wine with them sour grapes.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 18:53 UTC (Tue) by felixfix (subscriber, #242) [Link]
wine .... I think you meant whine. And he already has, and continues to do so.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 6:37 UTC (Tue) by sitaram (subscriber, #5959) [Link]
is this whine still going on? I admire his tenacity.
I've long ago started telling people to *stop* using or considering MySQL as far as possible, and to convert (to Pg, usually) if they can. I explain that because of the copyright assignment clause it was never a *true* open source project with proper community involvement in the code. So you now have one group that "owns" it but may or may not have the people to push it, and another than knows/claims to know the code best and is most suited to continue development/maintenance but doesn't own it.
None of this would have happened if the copyright assignment clause didn't exist. (And before any GNU guys start bashing me; I do make a distinction in my writings between assignment to a commercial entity whose future plans are quite unknown and a non-profit like FSF with stated goals for the far future).
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 8:09 UTC (Tue) by michaeljt (subscriber, #39183) [Link]
> I explain that because of the copyright assignment clause it was never a *true* open source project with proper community involvement in the code.
I may be slightly biased, as I work on such a project myself, but I don't see a problem. The person the copyright is assigned to can do what they want with the code themselves, but they can never take it away from other people (a.k.a the community). Monty can continue working on MySQL (I believe he is or was doing that, if he hasn't turned to full-time lobbying), he just can't make proprietary versions. If that point makes the software unfree you might consider informing Mr Stallman of the issue.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 9:08 UTC (Tue) by sitaram (subscriber, #5959) [Link]
Sorry, I didn't mean unfree, nor did I mean that *technically* it is not open source.
Since I'm all for reuse, even in arguments, I'm rehashing something I wrote when this first hit the wires in January.
----
I call this NQOSS ("not quite open source software") because such products have no (or very few) contributors from the outside world. MP have themselves said (see section 5.3 of http://openlife.cc/system/files/M.5529%20MP%20observation... ):
"[...] MySQL was almost fully developed by employees of MySQL Ab and later Sun's MySQL division."
In fact, from that statement I would even say MySQL has a closed source development model, even if the code is open source.
----
I hope that clarifies what I meant; sorry if it sounded different.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 9:44 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]
Open source definition doesn't require a diverse community. A company doing the majority of development still qualifies as "true" open source. Just not very healthy.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 9:48 UTC (Tue) by drag (subscriber, #31333) [Link]
Yeah.
If you look at the majority of open source projects the number of regular contributors is shockingly small.
For something as important as the entire Linux audio subsystem had as many as 2 or 3 developers for years.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 10:06 UTC (Tue) by robert_s (subscriber, #42402) [Link]
"Monty can continue working on MySQL (I believe he is or was doing that, if he hasn't turned to full-time lobbying), he just can't make proprietary versions."
Except, didn't MySQL always insist that the GPL covered their interface lib too, so supposedly you can't even write a proprietary application that uses MySQL?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 12:24 UTC (Tue) by hingo (guest, #14792) [Link] It's not a claim, you can download the clients and verify they are GPL too.This is the important bit to understand, and most of those who made a lot of noise in favor of Oracle last year completely missed the point of what a client-server architecture means. But it is not the end of the world. The client libraries are feasible to reimplement from scratch. In fact, the PHP client is already under the PHP language. Drizzle has reimplemented their own C and JDBC client library (BSD license), which can be extended to support the full MySQL protocol. So yes, Oracle/MySQL owns the app layer for proprietary apps (and GPLv3, curiously), but it is feasible to fix the problem by coding.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 12:25 UTC (Tue) by hingo (guest, #14792) [Link] Meant to say: "...under the PHP license"
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 20:38 UTC (Tue) by robert_s (subscriber, #42402) [Link]
"but it is feasible to fix the problem by coding."
Or by using a grownup database (ie postgres) instead.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 12:50 UTC (Tue) by cortana (subscriber, #24596) [Link]
Worse, they have claimed that the protocol itself is under the GPL, so anything that talks to their server must also be licensed under the GPL!
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 20:46 UTC (Tue) by HenrikH (guest, #31152) [Link]
Something that of course don't work, you cannot GPL a communications protocol. And as I have understood it, they nowdays say that the protocol is not covered by the GPL.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 23:17 UTC (Tue) by rahvin (subscriber, #16953) [Link]
That was how Monty and his ilk blackmailed everyone into the paid license. He always used the GPL as a threat, as in if you use the GPL version you have to GPL everything you do.
When Oracle was presented the opportunity to do the same to him he ran to the EU to protest the merger under the guise of FOSS but talked about proprietary software only in the appeal (and publicly denied doing so until the actual appeal was publicized showing his duplicity). In Monty's appeal to the EU he specifically requested that the EU force Oracle to relicense the mysql code-base as BSD. A decidedly bad move IMO that would only benefit private companies that want to take the source closed.
His single largest concern is that in 5 years Oracle will shut down the re-licensing and all that will be available is the GPL version. He then twists this argument to say that the loss of the pay to take private market will hurt the FOSS version even though they are only mildly connected since it was sold to SUN. Monty truly believes the only way to make money on FOSS is to fund it with private source licenses and he believes this is still a good market, as long as some huge company with little interest in this market doesn't buy it and take it away.
I fully expect Oracle to kill the relicensing after the 5 year agreement, and I fully expect the private mySQL market to die leaving only the GPL version behind. This will be better for everyone involved. With Oracle in possession of the copyright the entire private competition will be ensured a death that will stand as there is little value in selling the codebase.
All of the following was my opinion of course, but I think it's well founded from his public statements, and I will never trust someone that would lie about asking that the code be relicensed as BSD while claiming to represent the FOSS community interests publicly.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 22:00 UTC (Thu) by hingo (guest, #14792) [Link] In Monty's appeal to the EU he specifically requested that the EU force Oracle to relicense the mysql code-base as BSD.I wrote that appeal (together with Florian Müller, whom we hired as advisor). Since this is LWN, I would like to correct the above (elsewhere I ignore these comments). We have widely been accused of this in various reports throughout the blogosphere. The truth is we consistently pleaded for 1) divestiture of MySQL and 2) should Oracle get to keep MySQL, in a separate listing for various smaller details, such as pointing out the manual isn't open source or that MySQL is incompatible with GPL3. We never proposed to the EU a requirement for relicensing. The background to this false accusation was Oracle's original claim that since MySQL is open source, it doesn't really matter who owns it. Our reply to this was that given how the GPL and MySQL's business models work, Oracle's claim is not true at all. To illustrate a difference, we pointed out other Sun software under the Apache license and explained that for such software Oracle's claim would be (more) true. (Essentially we explained to the EU that GPL is not the same as public domain.) Even then we consistently made it clear such a license change would still not address many other anti-trust problems, and was not a preferred solution by us. The appeal you refer to can be found on Monty's blog. It's still perplexing to me that even if these accusations were wildly spread, nobody actually seems to read the paper. (Ok, so it's a bit long...)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 11:53 UTC (Wed) by marcH (subscriber, #57642) [Link]
I do not see how a communication protocol is legally different from an API. I have never seen the word "serialization" mentioned in legalese.
If you implement the MySQL protocol it is hard to claim that your work is not derived from MySQL.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 12:59 UTC (Wed) by anselm (subscriber, #2796) [Link] At least here in Germany (and probably in all of the EU) it is legal to implement a protocol (or, for that matter, API) if that is required for interoperability with another piece of software – regardless of how that software is licensed. You are even allowed to disassemble or otherwise reverse-engineer the relevant parts of the original software to find out how the protocol works, if the protocol is undocumented and the manufacturer of the software refuses to release appropriate specs. (Disclaimer: IANAL.)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 13:06 UTC (Wed) by michaeljt (subscriber, #39183) [Link]
> I do not see how a communication protocol is legally different from an API. I have never seen the word "serialization" mentioned in legalese.
My non-legal understanding is that these things are actually not always legally clearly defined. So if you find yourself in that uncomfortable position, you always have a certain chance that the judge will decide one way and a certain chance in the other.
I believe though that a good test of whether a piece of software is derived from a second is whether the main functionality of the first keeps on working in the absence of the second.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 16:12 UTC (Wed) by nye (guest, #51576) [Link]
>I believe though that a good test of whether a piece of software is derived from a second is whether the main functionality of the first keeps on working in the absence of the second.
I don't think that test is a good one, because a literal interpretation would lead to the conclusion that, eg, Firefox is a derivative work of my router's TCP/IP stack. Clearly a line must be drawn somewhere, presumably according to whether the second piece of software can be replaced with some other equivalent, but this line of reasoning leads us back to where we started.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 16:32 UTC (Wed) by michaeljt (subscriber, #39183) [Link]
> I don't think that test is a good one, because a literal interpretation would lead to the conclusion that, eg, Firefox is a derivative work of my router's TCP/IP stack.
Firefox works perfectly well here, although I most likely don't have your router's TCP/IP stack on my own router.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 17:02 UTC (Wed) by nye (guest, #51576) [Link]
As I said, you're drawing a line based on substitution of an equivalent.
Where do you put that line? Is it a derivative work if the substitution must be performed on the machine running the software in question? Does the copyright status of a program change if a shared library is replaced with a different one with a compatible API? How about if it's done years after the software was released? Is software A a derivative work of software B, until the point when a third party develops a drop-in substitute for B?
The test doesn't provide any clarification to any uncertain cases, because it is no less vague than the status quo.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 23:49 UTC (Wed) by marcH (subscriber, #57642) [Link]
> As I said, you're drawing a line based on substitution of an equivalent.
I don't know if this line is legally correct, but at least it is very clear cut.
> Where do you put that line?
"Based on substitution of an equivalent"
> Is it a derivative work if the substitution must be performed on the machine running the software in question?
Why would that make any difference?
> Does the copyright status of a program change if a shared library is replaced with a different one with a compatible API?
The test is not "if it is replaced" but "if it can be replaced".
> How about if it's done years after the software was released?
What matters is simply the restrictions on the API or on the protocol, since this is what allows equivalents to exist (even if they do not exist yet). As long as the API is the exclusive property of a company, this prevents any equivalent from ever being implemented.
> Is software A a derivative work of software B, until the point when a third party develops a drop-in substitute for B?
The drop-in substitute simply cannot be developed as long as the API is protected.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 11:51 UTC (Thu) by nye (guest, #51576) [Link]
>> As I said, you're drawing a line based on substitution of an equivalent.
>I don't know if this line is legally correct, but at least it is very clear cut.
Since I still don't understand your criteria for equivalency, I disagree.
>> Where do you put that line?
>"Based on substitution of an equivalent"
Which you haven't defined and, I contend, *could never* define with sufficient rigour.
>> Is it a derivative work if the substitution must be performed on the machine running the software in question?
>Why would that make any difference?
I'm not saying it would; I'm trying to understand your reasoning because it is unclear to me - your answer to the next part helps to clarify that though.
>> Does the copyright status of a program change if a shared library is replaced with a different one with a compatible API?
>The test is not "if it is replaced" but "if it can be replaced".
Everything *can* be replaced, in principle.
>> How about if it's done years after the software was released?
>What matters is simply the restrictions on the API or on the protocol, since this is what allows equivalents to exist (even if they do not exist yet). As long as the API is the exclusive property of a company, this prevents any equivalent from ever being implemented.
This makes no sense. You can always reimplement an API for the purposes of interoperability, barring patent problems. In most jurisdictions (not including the US AFAIK), you're even allowed to use reverse engineering to do so. If it weren't the case, something like Wine would have problems, and the PC clone explosion could never have happened.
>> Is software A a derivative work of software B, until the point when a third party develops a drop-in substitute for B?
>The drop-in substitute simply cannot be developed as long as the API is protected.
I'm not sure what you mean by 'protected', but this reasoning seems circular. You appear to be saying that if it can't be replaced, anything using it is a derivative work and hence it is 'protected', and if it's 'protected', then it can't be replaced.
On the flip side, you *also* seem to be saying that, for example, nobody needs to worry about OpenSSL linking exceptions based on the existence of an equivalent library (GNUtls) even if they *actually* incorporate OpenSSL.
I could release proprietary software which includes a GPL implementation of a known specification, and claim that the GPL component *could* be replaced, hence I'm not required to release the source for my part of the software. Is that your intention?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 14:02 UTC (Thu) by michaeljt (subscriber, #39183) [Link]
> Which you haven't defined and, I contend, *could never* define with sufficient rigour.
I'm getting the feeling that you are trying to get more out of my comment above than it will yield. Perhaps I expressed myself badly, but I was wanting to point out a test which, as far as I know, can sometimes be used to show that one work is not a derivative work of another (the thing with equivalents came along in later comments).
And as I also pointed out in that or an earlier posting, I don't think that these things do get defined rigourously in a legal context. Particularly in the context of software and the law, where specialised technical issues of necessity have to be interpreted by non-specialists, I believe there is always room for interpretation and pulling in aspects you hadn't thought of as a non-lawyer (or even as a lawyer) which can play against you when you thought you were doing tricky things in a watertight way.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 14:42 UTC (Thu) by nye (guest, #51576) [Link]
Well I got slightly confused by the fact that I'm talking to two people and hadn't noticed.
But yes, I think I probably did misunderstand the nature of your comment. Taking it as a starting point rather than a binary test does seem to be reasonable. I just don't like the thing with the equivalents since it muddies the water too far.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 23:19 UTC (Thu) by marcH (subscriber, #57642) [Link]
I take the blame for digressing in a number of random directions. Please let me focus back on the core question: does linking to or reimplementing the API of a GPL'ed library make your work "derived" from this library? I contend that a simple "substitution" test answer the above question, with reasonably good rigour. Thanks for ignoring the digressions in previous posts.
- if the API of this GPL library is an *original* design (e.g.: the MySQL API or protocol), then anything linking to or reimplementing it is by nature derived from it, and must be GPL'ed too. If MySQL had never existed your application linking to its API would have been obviously quite different. Same for any re-implementation of the MySQL API or protocols; they are derived from MySQL because they would have been quite different if MySQL had never existed.
- if the API of this GPL library is NOT an original design but implementing some existing and unrestricted standard (e.g. POSIX, TCP/IP,...), then anything linking to or re-implementing the GPL library is not derived from it; it derives straight from the standard. The "substitution" test means that you would have produced the exact same work even if the GPL library had never existed.
I do not pretend that this "substitution test" is the perfect definition of a derived work in the GPL sense. However I pretend it is a _simple and non-ambiguous_ definition.
> You can always reimplement an API for the purposes of interoperability, barring patent problems.
Please let's not consider interoperability here but focus on what is a derived work as defined in the GPL. Maybe some laws about interoperability conflict in one way or the other with the GPL? I dunno. It would be neither the first nor the last inconsistency in the legal field.
> this reasoning seems circular. You appear to be saying that if it can't be replaced, anything using it is a derivative work and hence it is 'protected', and if it's 'protected', then it can't be replaced.
It looks circular because it is the same thing. Just like: "he is alive because he is breathing", "he is breathing because he is alive".
> I could release proprietary software which includes [you meant: "links to", right?] a GPL implementation of a known specification, and claim that the GPL component *could* be replaced, hence I'm not required to release the source for my part of the software. Is that your intention?
This is already happening, otherwise there would not be much proprietary (and portable) software moving from Unix to Linux.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 23:26 UTC (Thu) by sfeam (subscriber, #2841) [Link] > You can always reimplement an API for the purposes of interoperability, barring patent problems.>>Please let's not consider interoperability here but focus on what is a derived work as defined in the GPL. This is a common mistake. The question of whether something is a derived work is not in the scope of the GPL, so what the GPL says is not relevant. Only after you have already decided for other reasons that a work is derivative, then you may [or may not] need to look at the GPL text to see what restrictions apply.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 13:35 UTC (Fri) by marcH (subscriber, #57642) [Link]
> > ... but focus on what is a derived work as defined in the GPL.
> This is a common mistake,...
Sorry, my bad. I meant: "focus on what is a derived work IN THE CONTEXT OF the GPL".
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 0:30 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]
by your definition anything that uses the windows API is a derivitive work of windows as it would not have been written if windows didn't exist.
things don't work that way.
I think that the FSF is wrong to the point of being activly harmful to try and claim that linking to a GPL library makes your work a derivitive of the library.
however since the FSF claims this, it opens the door for confusion when others try to claim similar things as well.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 13:02 UTC (Fri) by vadim (guest, #35271) [Link] I think that the FSF is wrong to the point of being activly harmful to try and claim that linking to a GPL library makes your work a derivitive of the library.Why would it be harmful? It's precisely the effect I want. If I didn't want it, I'd use the LGPL. If I use the GPL on a library, I fully intend the "linking to a GPL library makes your work a derivitive of the library" effect to happen. If somehow it turned out not to work that way, I wouldn't let it be, and would change the license on the library to achieve this effect anyway.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 13:20 UTC (Fri) by michaeljt (subscriber, #39183) [Link]
> by your definition anything that uses the windows API is a derivitive work of windows as it would not have been written if windows didn't exist.
[snip] > I think that the FSF is wrong to the point of being activly harmful to try and claim that linking to a GPL library makes your work a derivitive of the library.
Out of interest, what harm do you expect to come of this? I would have thougtht that Microsoft would not have in interest in trying to push this point of view, as it might discourage people from developing commercial software for Windows (and even if they did, not having done so for so many years would set a sort of precedent for exempting existing software).
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 13:36 UTC (Fri) by marcH (subscriber, #57642) [Link]
> by your definition anything that uses the windows API is a derivitive work of windows as it would not have been written if windows didn't exist.
Yes. And?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 20:23 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]
and if your work is a derivitive of another piece of work you are not allowed to distribute it without permission of the author of the work you are a derivitive of.
this would mean that if you wrote an app on windows you cannot distribute it without permission from microsoft.
that's just not how it works.
to answer the other questions here all at once. the reason I think that the FSF stance on interfacing with GPL code is bad is that by making that argument they confuse people on what is derivitive, and the definitions that many GPL supporters would like to use to protect their GPL code could be used by other companies just as readily to block GPL code on their systems.
to write an app for an iphone you must use the apple libraries, does that make your app a derivitive of those libraries? what if your app was origionally written for android and then ported to the iphone?
this is an extremely powerful tool to hand over to those companies that are interested in shutting down free software.
there's also a very hard problem determining where the boundry really exists.
you claim that if I make a C call to your library I am a derived work of your software. so If I use the openssl libraries I am a derived work of those libraries.
why am I not a derived work of your software if I write an expect script to achieve the same result through a human readable interface? if I get the exact same result as above, but by using the openssl command line tool how does that change me to not be a derived work?
as I see it the only sane answer (and the one I think the law will eventually settle on in another decade or two) is that any use through a published interface does not constitute a derived work. It doesn't matter if that use is via a function call after the OS links your code to my code, via a network call, or via screen-scraping the human readable UI of your application.
yes, this will make the GPL and LGPL essentially the same, but it also cuts the legs out from under so many proprietary abuses of copyright that I think it's well worth it.
David Lang
Monty appeals Oracle's Sun merger (The Register) Posted Jul 14, 2010 11:55 UTC (Wed) by marcH (subscriber, #57642) [Link]
> and if your work is a derivitive of another piece of work you are not allowed to distribute it without permission of the author of the work you are a derivitive of. this would mean that if you wrote an app on windows you cannot distribute it without permission from microsoft.
That's why Microsoft licences grant you this permission, since Microsoft is releasing libraries not to be contemplated but to actually use them.
No inconsistency here really.
> yes, this will make the GPL and LGPL essentially the same,
This could be the best summary of your opinion, thanks for that.
> but it also cuts the legs out from under so many proprietary abuses of copyright that I think it's well worth it.
The GPL has been explicitely designed in reaction to abusive proprietary licences, so it is only fair that it tries to (ab)use the legal system(s) as well. I am sure the FSF would be delighted if some court somewhere rules that the GPL and the LGPL are the same, thereby having similar consequences on proprietary software.
I think you are overrating the influence of the GPL on legal system(s). Proprietary software licences are routinely full of abusive clauses just in case they might be valid somewhere, and they really do not need any lesson from the FSF for that. The GPL does not get to decide what a "derived work" is but it is only fair for it to be written with the most pessimistic assumption in mind. It might be true somewhere.
> you claim that if I make a C call to your library I am a derived work of your software
*I* am not. Just like the GPL, I think it is just better to consider this pessimistically wide definition of a derived work and be prepared for it. Especially since there is no inconsistency of any kind in it. Precisely because it is the simplest definition.
And just like you, I obviously wish to be wrong.
> why am I not a derived work of your software if I write an expect script to achieve the same result through a human readable interface?
You would be the same (have fun defining "human readable")
Monty appeals Oracle's Sun merger (The Register) Posted Jul 14, 2010 23:58 UTC (Wed) by dlang (✭ supporter ✭, #313) [Link]
as for a definition of human readable, as an extreme situation, what if I make a program that can read the screen of microsoft outlook, clicking on the appropriate locations, etc?
or an application that extracts your banking data from www.bankofamerica.com (with you providing the application your password)
these are very obvious cases where the interface is designed to be human readable.
in neither case is my work a derived work of the thing I am talking to
Monty appeals Oracle's Sun merger (The Register) Posted Jul 15, 2010 10:28 UTC (Thu) by marcH (subscriber, #57642) [Link]
I suspect courts are generally not interested in technical details like: "Is this interface human-readable?". They would rather just look at things at higher level, like: "Is this *specific* piece of software absolutely required for this other piece to work?"
Monty appeals Oracle's Sun merger (The Register) Posted Jul 15, 2010 10:44 UTC (Thu) by dlang (✭ supporter ✭, #313) [Link]
actually, being a derivitive for copyright purposes has absolutly nothing to do with "Is this *specific* piece of software absolutely required for this other piece to work?"
if you didn't copy a single line of code from the other software, there is absolutly no possible way that you can be derived from that code.
not everything that is written is protected by copyright.
So, even if you did copy code from the other software, you may still not be derived from it if the code that you copied was not protected by copyright. This is where things get harder to hash out as many portions of header files are definitely not protected by copyright.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 16, 2010 8:48 UTC (Fri) by paulj (subscriber, #341) [Link]
I'm not sure that's true. E.g. havn't there been cases where fanfics have been found to be derived from the original works? No literal copying there...
Monty appeals Oracle's Sun merger (The Register) Posted Jul 16, 2010 12:07 UTC (Fri) by anselm (subscriber, #2796) [Link] As long as you don't copy recognisable passages verbatim from the original book then you ought to be technically in the clear. You might not come across as terribly original, but that's that. Ever wonder why publishers try to trademark the names of their famous characters (think »Mr Spock™«)? Right, because the names or characters can't be copyrighted by themselves, as a concept – only the text of the stories they appear in can. (Disclaimer: IANAL.)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 16, 2010 12:44 UTC (Fri) by cortana (subscriber, #24596) [Link]
This is the traditional time to bring up <http://www.plotpatents.com/> and poke fun at it. :)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 16, 2010 21:14 UTC (Fri) by mp (subscriber, #5615) [Link] From a comment, apparently by a lawyer, posted here:In short, substantial copying of exact expression is one but not the only way of infringing copyright, under US law, UK law, and the Berne Convention.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 17, 2010 19:35 UTC (Sat) by nix (subscriber, #2304) [Link]
Yes, Charlie Petit is definitely a lawyer. Very definitely.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 15:35 UTC (Fri) by vonbrand (subscriber, #4458) [Link] Same problem... POSIX was codification of Unix API. So before POSIX it was protected, afterwards not? Who allowed IEEE to create POSIX then? If I go and write a "standards document" about MySQ"L's API or protocol, does that by itself make them "not protected"? If it isn't just me, but a few dozen random hackers on the 'net? The LWN readers? IETF, ACM, ISO?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 16:24 UTC (Fri) by marcH (subscriber, #57642) [Link]
> POSIX was codification of Unix API.
The Unix API was not protected by the GPL, so this is not a relevant example.
> If I go and write a "standards document" about MySQ"L's API or protocol,...
Documenting something does not magically free it from the GPL.
A standard is not just documentation, it is also a permission to use it.
> If it isn't just me, but a few dozen random hackers on the 'net?
This is not about you, it's about MySQL deciding to free its API
An API or a protocol is just a (tiny & special) piece of source
You can argue on the other hand that an API is "too
Monty appeals Oracle's Sun merger (The Register) Posted Jul 9, 2010 17:40 UTC (Fri) by nix (subscriber, #2304) [Link]
Whether something is or is not a derived work has *nothing to do with the GPL at all*. That's like saying that, oh yes, of course a burglar should be allowed to define what constitutes theft, or a fraudster what constitutes fraud (oh, wait). Legal documents like licenses do not get to choose whether they apply to other works or not: that decision is up to other parts of the law (and, ultimately, the human beings in the legal system).
So Horst's POSIX point is entirely apposite, because the same rules would apply to things being a derived work of the POSIX API document as apply to programs being a derived work of the GPL. (These rules are designed for books and performance art, mostly, so applying them to software is arcane and ridiculous, and I'm glad I don't know all the rules because I think my brain would dribble out of my ears. Those rules differ between jurisdictions anyway.)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 14, 2010 3:29 UTC (Wed) by vonbrand (subscriber, #4458) [Link] My further point is that some description of software or protocols in a standards (or other) document can't magically make what was some "protected sauce" into "open for all".
Monty appeals Oracle's Sun merger (The Register) Posted Jul 14, 2010 14:12 UTC (Wed) by marcH (subscriber, #57642) [Link]
> Legal documents like licenses do not get to choose whether they apply to other works or not
I am not discussing that. I am only considering when this decision has already been made by some legal system somewhere in the world. I mean, I am only considering the case where some legal system has decided that yes, linking to an API makes a derived work.
> So Horst's POSIX point is entirely apposite, because the same rules would apply to things being a derived work of the POSIX API document as apply to programs being a derived work of the GPL.
So you are basically saying that writing "derived work" in any licence of any kind is just pointless, correct? Any attempt at controlling works derived from yours is futile: in whatever place you live it's all defined in the Law.
I doubt it.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 14:01 UTC (Wed) by njs (guest, #40338) [Link]
No, but copyright covers creative expression, not function. So if the only possible way to achieve a particular function (e.g., talking to a MySQL server) is by producing a particular stream of bytes, then those bytes cannot be expressive, since creative expression necessarily involves choice.
Or at least that's one legal theory current in the USA. YJudgeMV.
(See, e.g., http://www.chillingeffects.org/reverse/faq.cgi#QID199)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 14:39 UTC (Wed) by mjg59 (subscriber, #23239) [Link]
This argument leads to things like implementing recent versions of the SMB wire protocol creating a derivative work of Windows.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 16:24 UTC (Wed) by foom (subscriber, #14868) [Link]
Or to things like claiming that if you dlopen("libreadline.so.5") and dlsym() yourself some functions in it, that your software is a derived work of readline, and thus falls under the requirements of the GPL.
Of course, clearly that's utterly ridiculous and nobody would ever claim that, right?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 16:31 UTC (Wed) by cortana (subscriber, #24596) [Link]
Hey, when I link myproprietaryapp.o against libreadline.so.5 I'm 'merely aggregating' the two pieces of software together, and can therefore ignore the requirements of the GPL. :)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 17:59 UTC (Wed) by foom (subscriber, #14868) [Link]
Well, you're not *even* "aggregating" them, unless you distribute them together. You're just talking over a well-defined communications channel to another piece of software that's already on the user's machine.
So yea, the GPL is clearly irrelevant in that case, right?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 18:24 UTC (Wed) by nix (subscriber, #2304) [Link]
Don't such programs generally include readline header files as well? Textual inclusion of a GPLed work...
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 20:56 UTC (Wed) by dlang (✭ supporter ✭, #313) [Link]
header files may not be copywriteable
there are quite a few things that cannot be covered by copywrite, and header files (being lists of definitions of variables and functions that are necessary to interoperate with the software) generally fall into the definition of things that can't be covered.
this is more than just theory, it's been tested in court.
in the lexmark printer cartridge case the court rules that the entire executable of the code in the cartridge was not covered by copyright as it was necessary to duplicate it bit for bit in clone cartridges to have them interoperate with the printers.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 23:17 UTC (Wed) by nix (subscriber, #2304) [Link]
But it is surely not necessary to duplicate header files bit by bit in order to interoperate with libraries. Not in C. You need a couple of structure definitions, but the prototypes, while a good idea, are not mandatory. (But IANAL and I know lawyers don't think this way...)
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 1:42 UTC (Thu) by dlang (✭ supporter ✭, #313) [Link]
the thing is that if there is nothing in the file that qualifies for copyright protection there's no reason not to just copy the entire file.
you would have to make the argument that somehow the arrangement of the data in the file was a creative endevor, and that's a pretty hard thing to do when the order of the data in the file really doesn't matter.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 2:00 UTC (Thu) by foom (subscriber, #14868) [Link]
The original file has comments.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 12:55 UTC (Thu) by nix (subscriber, #2304) [Link]
Exactly. It also has a lot of other nonfunctional entities: the names of identifiers other than those corresponding to exported symbols in the library, for example.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 19:57 UTC (Thu) by dlang (✭ supporter ✭, #313) [Link]
but do any of those entities contain sufficient original content to qualify for copyright protection?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 21:05 UTC (Wed) by foom (subscriber, #14868) [Link]
Well, let's say I don't include the readline header, but instead I extracted all the function, variable, macro, and struct declarations from it into a new header "readline-interface.h", which I license in the public domain.
Clearly I'm allowed to do that, because there is no creative expression in those definitions: they're a simple statement of fact of how to interoperate with the readline library. Right?
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 23:18 UTC (Wed) by nix (subscriber, #2304) [Link]
I suspect that that would work, certainly if there was a cleanroom in the middle, so that all that passed across was a description of the prototypes and structure definitions. But the fact is that nobody *does* this.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 8, 2010 21:49 UTC (Thu) by hingo (guest, #14792) [Link] Just in case anyone wants to create a MySQL compatible client, I may as well point out that current MySQL owner does not claim the protocol to be GPL (but Monty indeed did). http://bugs.mysql.com/bug.php?id=52107As a practical example, the native PHP driver for MySQL is under the PHP license, and nobody has ever claimed that all of PHP magically is a derivative work of MySQL and thus GPL.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 13:19 UTC (Tue) by anselm (subscriber, #2796) [Link] Except, didn't MySQL always insist that the GPL covered their interface lib too, so supposedly you can't even write a proprietary application that uses MySQL? Even if that is the case, I don't see why that should prevent anybody from implementing their own (free or proprietary) library that talks the MySQL wire protocol. To be extra safe, have somebody write up a spec from the MySQL source and somebody else code the alternative implementation.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 7, 2010 11:59 UTC (Wed) by marcH (subscriber, #57642) [Link]
> MySQL always insist that the GPL covered their interface lib too, so supposedly you can't even write a proprietary application that uses MySQL?
Even if you are not using any MySQL API but only standard APIs? (JDBC, SQL,...)
This assumes that your proprietary application is extremely portable, so please consider this as a purely theoretical question.
Monty appeals Oracle's Sun merger (The Register) Posted Jul 6, 2010 8:12 UTC (Tue) by drag (subscriber, #31333) [Link]
> I explain that because of the copyright assignment clause it was never a *true* open source project with proper community involvement in the code.
It was, and is, a true open source project. The copyright assignment is understandably annoying and I can see how that would be unattractive.
But you have to realize the end effect of everybody owning a small copyrighted portion of MySQL would be exactly the same as if Oracle decided to make it closed source... which is you have a GPL'd project were none of the major players have control over the license. This is exactly what you have with the Linux kernel.
If Oracle becomes too irritating then it's not a big deal to move to something like MariaDB or one of the other forks that were made in preparation of Oracle turing into a total dick (which so far has not happened). I don't see why it would be necessary at all to move to postgresql if that was the sort of a developer is worried about. It's just paranoia and would require a lot of work that, I expect, is otherwise unnecessary.
|
|||||||||||||