Re: copyrightable APIs? (was RE: namespace protection compatiblewit
On Sun, 22 Apr 2001, Angelo Schneider wrote: phil hunt wrote: On Fri, 20 Apr 2001, Angelo Schneider wrote: Hi! In Europe APIs are not copyright able. No idea about the US. However if you publich them in a book, the book of course is copyrighted. However you can not prevent anyone to write a software against a given API. Same is true for data formats. (In Europe dataformats e.g. a flat file format for a word processor are not copyright able) This will change under the new EU copyright law, where it will be illegal to decrypt any encrypted file format (e.g. DVD) without the copyright holder's permission. Thats a misunderstanding. Just for simplicity lets talk about MS Word. The file format is proprietary. I do not know it. However I'm free to analyze Word files and discover it. I'm free to write Word files after I have discovered how they look like. Now you have two possibilities: Encrypt the whole word file - decryption is illegal if the decrypter has to asume the file contains private data(right of private speech is broken). The EU copyright directive says *nothing* AFAIK about private speech, or rights thereto. Encrypt the content of the file BEFORE you write it in Word format (e.g. by keeping the paragraph structure and encrypting each paragraph) - decryption is covered by the new law and considered illegal (the law has not passed the houses so far). The EU copyright directive (which you are right isn't law yet) makes it illegal to circumvent a technological measure. A technological measure includes any encryption put on a file. Both points don't touch the fact that the API or the format it self is not copyrightable, how should one be able to WRITE a DVD then? My understanding is that it is illegal to write an encoded DVD in the USA without the permission of the DVD-CCA. -- * Phil Hunt * An unforseen issue has arisen with your computer. Don't worry your silly little head about what has gone wrong; here's a pretty animation of a paperclip to look at instead. -- Windows2007 error message
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
Hi all! Rod Dixon wrote: Those are very good thoughts, if I may say so. Rod On Fri, 20 Apr 2001, Chloe Hoffman wrote: I am not sure I see how 102(b) should exclude APIs from copyrightable subject matter as an absolute matter. Surely some aspects of an API may fail because of various doctrines such as merger, scenes a faire, etc. (viz. sqrt()) but I am not sure I see how a full set of APIs should be excluded per se. I find it hard to distinguish an API from a "computer program" - if APIs fail under 102(b) then shouldn't computer programs in An API is not THE program. A POSIX compliant kernal (or his calling interface) all suport the same API, POSIX. The standard C library and the glibc have a lot of "functions" in common. The command line options of the command "ls" or for simplicity a command like "rm" are mre or less the same in the same UNIX family. THAT is an API. Of course one can write a replacement for glibc exposing the same API. Of course one can write a kernal which has a calling interface which is identical to that of POSIX. Of course one can write a replacement for "ls" or "rm" which accept the same command line options (and serve the same purpose). An API is in my sence a LANGUAGE. Take it as an mathematical language if you like, a language can not be copyrighted. The other posts I saw regarding this thread (about Adope, Psotscript and PDF, and Sun/Microsoft JAVA) are in my opinion Trademark issues. The same is true if you go and call a kernal POSIX compliant. I don't know if POSIX is a trademark. The only thing which is in general copyrightable, is CONTENT. The only thing which is in general(exceptions exist e.g. in the US) patent able is a PROCESS in conjunction whith the artifact which is created by performing it. (Same process for different purpose is not affected, same artifact created in a different way is not affected - except that artifact is copyrighted) The only thing wich can get trademarked are (artificial) names used in public (business) affairs (where the names reffer to an artifact or an business). Well, I'm not a lawyer and I simplificated it a bit. (Also I'm from germany, a lot of stuff is different here than in the rest of the world, outside europe) However I work in copyright relevant areas for 10 years now Finaly: JAVA is a trademark, so if GNU CLASSPATH would call it self JAVA, than there would be trouble. Regards! Angelo general fail also because they comprise an idea, process, method, etc.? I see both as expressions, not the idea themselves. I think the tougher issue is infringement/derivative works (leave alone implied/express licenses, estoppels, etc.). Just some thoughts From: Rod Dixon To: Angelo Schneider CC: , Subject: Re: copyrightable APIs? (was RE: namespace protection compatible wit Date: Fri, 20 Apr 2001 14:12:13 -0400 (EDT) This is the issue I was hinting at. I do not believe that as a general matter that APIs should be copyrightable under U.S. copyright law since section 102(b) of the Copyright Act should exclude APIs from copyright subject matter. Having said that, I admit the issue seems unresolved since both Microsoft and Sun Microsystems are two well known developers who claim copyright interests in APIs; Microsoft for Windows, and Sun for Java. Rod On Fri, 20 Apr 2001, Angelo Schneider wrote: Hi! In Europe APIs are not "copyright able". No idea about the US. However if you publich them in a book, the book of course is copyrighted. However you can not prevent anyone to write a software against a given API. Same is true for data formats. (In Europe dataformats e.g. a flat file format for a word processor are not copyright able) Regards, Angelo Forrest J Cavalier III wrote: -- Von: Forrest J Cavalier III[SMTP:[EMAIL PROTECTED]] Gesendet: Freitag, 20. April 2001 13:50:06 An: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Betreff: copyrightable APIs? (was RE: namespace protection compatible wit Diese Nachricht wurde automatisch von einer Regel weitergeleitet. How can you copyright an API? Isn't it simply a collection of facts? Perhaps you could copyright the formal parameter names, and certainly the documentation in a header file. But the facts of function name, return type(s) parameter type(s) are just facts. There is no creative expression involved. Forrest J. Cavalier III, Mib Software Voice 570-992-8824 http://www.rocketaware.com/ has over 30,000 links to source, libraries, functions, applications, and documentation. -- Angelo Schneider OOAD/UML [EMAIL PROTECTED] Putlitzstr. 24 Patterns/FrameWorks Fon: +49 721 9812465 76137 Karlsruhe C++/JAVA Fax:
RE: copyrightable APIs? (was RE: namespace protection compatiblewit
-Original Message- From: Lawrence E. Rosen [mailto:[EMAIL PROTECTED]] Sent: Friday, April 20, 2001 11:24 PM To: [EMAIL PROTECTED] Subject: RE: copyrightable APIs? (was RE: namespace protection Finally, one CAN use trademark law -- with all its strengths and weaknesses -- to prevent third parties from applying your trademark to their goods, or to prevent third parties from applying your certification mark to their incompatible goods. For example, legitimate trademarks or certification marks for "Java" or "Windows," (if they exist; I haven't searched!) could be used to prevent people from forking Java and Windows APIs and still calling those imprecise implementations "Java" or "Windows." Hello Larry, Good point! Microsoft fought like hell to obtain a federally registered trademark on "Microsoft Windows." After initial failure, there persistence eventually paid off. Sun claims to have a common law trademark on various iterations of "Java," but no federal registration has been obtained. Of course, Sun used its trademark and license as a basis to file suit against Microsoft for "implementing" the Java API in an incompatible manner. Although it looks as if the terms of the settlement agreement allow Microsoft to continue to support (and, perhaps, promote) third party Java application development on the Windows platform that is incompatible with the Sun API, - - but without use of the common law java trademark - - it's unclear from the terms whether Microsoft agreed it was paying to license a copyright interest in Sun's Java API. We know Microsoft did not pay to license the trademark. I hope that other attorneys on this discussion list will help me evaluate whether a court challenge to the use of the copyright law to protect APIs is likely to succeed [SNIP] I think this is a difficult question to answer in the abstract. Microsoft seems to have raised the argument indirectly (and quite cautiously) in its antitrust litigation. Judge Jackson did not give the argument much response, but during oral argument before the D.C. Circuit, one could have drawn an inference that the D.C. Circuit might actually be sympathetic to a clearly presented argument. Hence, I think a court challenge to the use of the copyright law to protect APIs would be a gamble. It might be better to expect that courts will apply the abstraction/filtration test or the idea/expression dichotomy in cases claiming infringement of the copyright in a set of APIs in a manner that filters out those claims or flatly rejects them. One final note: one might argue that such a use of copyright constitutes copyright misuse, if faced with a claim of infringement. I do not know whether Microsoft raised that defense in its answer to the Sun lawsuit. The downside to this argument is that copyright misuse is a judicial doctrine without a statutory basis or explicit support from the Supremes. Hence, many courts reject the defense entirely or limit it to "antitrust-like" issues. Rod Rod Dixon Visiting Assistant Professor of Law Rutgers University Law School - Camden www.cyberspaces.org [EMAIL PROTECTED]
RE: copyrightable APIs? (was RE: namespace protection compatiblewit
On Fri, 20 Apr 2001, Lawrence E. Rosen wrote: Even if a company were to argue successfully that its API is *both* expressive and substantive, and thus protectible as copyrightable subject matter, I would argue that access to the API for the purpose of preparing independent (compatible or incompatible) software, even including making copies, is still allowed under the fair use provisions of the copyright act. As the court held in Sega v. Accolade, 977 F.2d 1510, 1521-1524 (9th Cir. 1992), in analyzing the four factors justifying a fair use defense: Compelling. If not "ironclad", this does appear to be the decision I was looking to be able to cite regarding the enforceability of copyright of an API over implementations. Thanks Larry; I'm now done with this topic. =) Brian
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
On Fri, 20 Apr 2001, Angelo Schneider wrote: Hi! In Europe APIs are not "copyright able". No idea about the US. However if you publich them in a book, the book of course is copyrighted. However you can not prevent anyone to write a software against a given API. Same is true for data formats. (In Europe dataformats e.g. a flat file format for a word processor are not copyright able) This will change under the new EU copyright law, where it will be illegal to decrypt any encrypted file format (e.g. DVD) without the copyright holder's permission. -- * Phil Hunt * "An unforseen issue has arisen with your computer. Don't worry your silly little head about what has gone wrong; here's a pretty animation of a paperclip to look at instead." -- Windows2007 error message
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
This is the issue I was hinting at. I do not believe that as a general matter that APIs should be copyrightable under U.S. copyright law since section 102(b) of the Copyright Act should exclude APIs from copyright subject matter. Having said that, I admit the issue seems unresolved since both Microsoft and Sun Microsystems are two well known developers who claim copyright interests in APIs; Microsoft for Windows, and Sun for Java. Rod On Fri, 20 Apr 2001, Angelo Schneider wrote: Hi! In Europe APIs are not "copyright able". No idea about the US. However if you publich them in a book, the book of course is copyrighted. However you can not prevent anyone to write a software against a given API. Same is true for data formats. (In Europe dataformats e.g. a flat file format for a word processor are not copyright able) Regards, Angelo Forrest J Cavalier III wrote: -- Von: Forrest J Cavalier III[SMTP:[EMAIL PROTECTED]] Gesendet: Freitag, 20. April 2001 13:50:06 An: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Betreff: copyrightable APIs? (was RE: namespace protection compatible wit Diese Nachricht wurde automatisch von einer Regel weitergeleitet. How can you copyright an API? Isn't it simply a collection of facts? Perhaps you could copyright the formal parameter names, and certainly the documentation in a header file. But the facts of function name, return type(s) parameter type(s) are just facts. There is no creative expression involved. Forrest J. Cavalier III, Mib Software Voice 570-992-8824 http://www.rocketaware.com/ has over 30,000 links to source, libraries, functions, applications, and documentation. -- Angelo Schneider OOAD/UML [EMAIL PROTECTED] Putlitzstr. 24 Patterns/FrameWorks Fon: +49 721 9812465 76137 Karlsruhe C++/JAVAFax: +49 721 9812467
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
I doubt whether we will resolve the copyrightability question. I think the better view is that an API is not copyrightable subject matter. I also think that viewing an API as such better serves the purposes of copyright law. Even so, I agree that the more important question is if you assume that an API *IS* copyrightable, what next? The answer is that programmers have an implied license to copy. Of course, that does not mean a programmer ought to be able to copy an entire API for a given programming environment, but it does mean that most copies of an API function or routine do not exceed the implied license. (BTW, not to add to any confusion, but my use of the term copy is in the strict copyright sense that applies to software). Rod On Fri, 20 Apr 2001, Chloe Hoffman wrote: In my view, an API is as much a collection of facts as your original message, as Stephen King's latest novel, etc. I think in most cases an API involves creative expression or at least some selection, arrangement or coordination of function names, parameter type(s) and return type(s) (of course I am not talking about the simple abstract concept of an API; I am talking about a set of developed APIs). Surely if an API is just one function then you have a de minimis problem. But let's take the Java API. Taking U.S. law as an example, I would think that after you take whatever material (functions, return types, parameter types, parameter names, etc.) that is not copyrightable (by virtue of, for example, the merger doctrine(the idea and expression merged into one and there is no other way of expressing it), the scenes a faire doctrine (only so many ways of expressing the idea) and being in the public domain) there would be a great deal of material left over that involved creative expression or at least serious selection, coordination, or arrangement. For copyright to attach only minimal originality is needed. I can't see the argument flying that the Java API is like a purely alphabetical white pages. I think the real question is not whether an API is copyrightable but how an API is infringed and what is a derivative work of an API. From: "Forrest J Cavalier III" Reply-To: [EMAIL PROTECTED] To: CC: [EMAIL PROTECTED] Subject: copyrightable APIs? (was RE: namespace protection compatible wit Date: Fri, 20 Apr 2001 07:50:06 -0400 How can you copyright an API? Isn't it simply a collection of facts? Perhaps you could copyright the formal parameter names, and certainly the documentation in a header file. But the facts of function name, return type(s) parameter type(s) are just facts. There is no creative expression involved. Forrest J. Cavalier III, Mib Software Voice 570-992-8824 http://www.rocketaware.com/ has over 30,000 links to source, libraries, functions, applications, and documentation. Get your FREE download of MSN Explorer at http://explorer.msn.com
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
Those are very good thoughts, if I may say so. Rod On Fri, 20 Apr 2001, Chloe Hoffman wrote: I am not sure I see how 102(b) should exclude APIs from copyrightable subject matter as an absolute matter. Surely some aspects of an API may fail because of various doctrines such as merger, scenes a faire, etc. (viz. sqrt()) but I am not sure I see how a full set of APIs should be excluded per se. I find it hard to distinguish an API from a "computer program" - if APIs fail under 102(b) then shouldn't computer programs in general fail also because they comprise an idea, process, method, etc.? I see both as expressions, not the idea themselves. I think the tougher issue is infringement/derivative works (leave alone implied/express licenses, estoppels, etc.). Just some thoughts From: Rod Dixon To: Angelo Schneider CC: , Subject: Re: copyrightable APIs? (was RE: namespace protection compatible wit Date: Fri, 20 Apr 2001 14:12:13 -0400 (EDT) This is the issue I was hinting at. I do not believe that as a general matter that APIs should be copyrightable under U.S. copyright law since section 102(b) of the Copyright Act should exclude APIs from copyright subject matter. Having said that, I admit the issue seems unresolved since both Microsoft and Sun Microsystems are two well known developers who claim copyright interests in APIs; Microsoft for Windows, and Sun for Java. Rod On Fri, 20 Apr 2001, Angelo Schneider wrote: Hi! In Europe APIs are not "copyright able". No idea about the US. However if you publich them in a book, the book of course is copyrighted. However you can not prevent anyone to write a software against a given API. Same is true for data formats. (In Europe dataformats e.g. a flat file format for a word processor are not copyright able) Regards, Angelo Forrest J Cavalier III wrote: -- Von: Forrest J Cavalier III[SMTP:[EMAIL PROTECTED]] Gesendet: Freitag, 20. April 2001 13:50:06 An: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Betreff: copyrightable APIs? (was RE: namespace protection compatible wit Diese Nachricht wurde automatisch von einer Regel weitergeleitet. How can you copyright an API? Isn't it simply a collection of facts? Perhaps you could copyright the formal parameter names, and certainly the documentation in a header file. But the facts of function name, return type(s) parameter type(s) are just facts. There is no creative expression involved. Forrest J. Cavalier III, Mib Software Voice 570-992-8824 http://www.rocketaware.com/ has over 30,000 links to source, libraries, functions, applications, and documentation. -- Angelo Schneider OOAD/UML [EMAIL PROTECTED] Putlitzstr. 24 Patterns/FrameWorks Fon: +49 721 9812465 76137 Karlsruhe C++/JAVA Fax: +49 721 9812467 Get your FREE download of MSN Explorer at http://explorer.msn.com
Re: copyrightable APIs? (was RE: namespace protection compatiblewit
On Fri, 20 Apr 2001, Chloe Hoffman wrote: Do you have any basis for the "better" view? Also, how does it better serve the purposes of copyright? Well, I said I *THINK* the better view is... In other words, I was expressing an opinion. The reason why I think it is the better view is because it serves the purposes of copyright NOT to view APIs as copyrightable subject matter, generally. (There might be some exceptions, but i cannot think of any). I cannot answer the "how" part in an e-mail message, but i could point you to a law review article, if you are interested. What happened to protection for authors for limited times? I am glad you asked the question. Authors who create copyrightable works gain the protection, those who do not don't (not even for a limited time). What we might disagree on is what is the constitutional purpose of copyright. For me, the constitution clearly sets out that the purpose of the congressional power to grant copyright interests is to *promote the progress of science and art.* What better way to do so than to let programmers develop works that ARE subject to copyright protection by ensuring free and open access to APIs (ideas), which constitute processes, ideas and methods outside the scope of copyright? Surely non-copyrightable APIs are good for the 'open source' software (or are they?) but I am not sure if it is better as a general proposition. I disagree. (see above) Also, I am not sure I see the basis for the implied license to copy (or perhaps even the more important right, to create a derivative work) in all cases. Good point here. As I mentioned earlier, some developers like Sun and Microsoft can weaken or disturb the implied license argument by explicitly stating to the contrary. However, merely asserting a copyright interest in an API is not sufficient. Indeed, the implied license argument assumes that such an assertion might be made. As for derivative work, I am unsure how that changes the analysis. The assumption is that at least one exclusive right has been infringed unless there is an implied license. If your position is that the creation of a derivative work must be viewed as outside the scope of the implied license, then we are in very cloudy territory. Certainly, we would not want to say that assuming an API is copyrightable, that any call to an API function results in a derivative work. Would we? There might be a strong disincentive for anyone who might own the copyright in an API to make that claim. Wouldn't you say so? - Rod Surely it is not in the interest of most software vendors to prevent copying/preparing derivative works of API (network effects and all that) but that does not necessarily mean those vendors rights are foregone. In many proprietary software cases no express rights have been given (contra see the license provisions in the Java APIs) nor could I see any such license being implied as necessary to the "use" of the licensed software (absent ambiguous or express language in that software's license). From: Rod Dixon To: Chloe Hoffman CC: , Subject: Re: copyrightable APIs? (was RE: namespace protection compatible wit Date: Fri, 20 Apr 2001 14:34:38 -0400 (EDT) I doubt whether we will resolve the copyrightability question. I think the better view is that an API is not copyrightable subject matter. I also think that viewing an API as such better serves the purposes of copyright law. Even so, I agree that the more important question is if you assume that an API *IS* copyrightable, what next? The answer is that programmers have an implied license to copy. Of course, that does not mean a programmer ought to be able to copy an entire API for a given programming environment, but it does mean that most copies of an API function or routine do not exceed the implied license. (BTW, not to add to any confusion, but my use of the term copy is in the strict copyright sense that applies to software). Rod On Fri, 20 Apr 2001, Chloe Hoffman wrote: In my view, an API is as much a collection of facts as your original message, as Stephen King's latest novel, etc. I think in most cases an API involves creative expression or at least some selection, arrangement or coordination of function names, parameter type(s) and return type(s) (of course I am not talking about the simple abstract concept of an API; I am talking about a set of developed APIs). Surely if an API is just one function then you have a de minimis problem. But let's take the Java API. Taking U.S. law as an example, I would think that after you take whatever material (functions, return types, parameter types, parameter names, etc.) that is not copyrightable (by virtue of, for example, the merger doctrine(the idea and expression merged into one and there is no other way of expressing it), the scenes a faire doctrine (only so many ways of expressing the idea) and being in the public
RE: copyrightable APIs? (was RE: namespace protection compatiblewit
I have a slightly different question about API's and copyrights. Suppose one has an API that acts as a specification for access to a library and perhaps a sample implementation. I dont care if someone creates another implementation of the API, in fact I want to encourage other people to do so. What I do not want is for someone to modify the API and therefore the specification of the libraries behavior. What is the role of copyrights in achieving this. I think that's another way of phrasing my original question. =) However you phrase the question, the answer is the same: Copyright protects expression and not the underlying ideas. You can prevent someone from copying your specification, creating derivative works of it, and distributing same. But you cannot prevent someone from implementing something independently (without copying!) after they read your specification. Nor, obviously, can you prevent someone from implementing something *different* from that which you describe in your specification. Even if a company were to argue successfully that its API is *both* expressive and substantive, and thus protectible as copyrightable subject matter, I would argue that access to the API for the purpose of preparing independent (compatible or incompatible) software, even including making copies, is still allowed under the fair use provisions of the copyright act. As the court held in Sega v. Accolade, 977 F.2d 1510, 1521-1524 (9th Cir. 1992), in analyzing the four factors justifying a fair use defense: (1) Copying is permitted "in order to discover the functional requirements for compatibility" for a "legitimate, essentially non-exploitative purpose," especially where there is a "public benefit" (in that case, "an increase in the number of independently designed ... programs"); [this is the "purpose or character of use" factor] (2) "To the extent that a work is functional or factual," or where copying of "expressive elements of the work ... 'must necessarily be used as incident to' expression of the underlying ideas, functional concepts, or facts," copying is permitted; [this is the "nature of the copyrighted work" factor] (3) "Where the ultimate (as opposed to direct) use" of the copied material "is limited," even the copying of the entire copyrighted work is permitted; [this is the amount and substantiality of the use" factor] (4) even where the copying is for the purpose of competing with the original copyright holder, "an attempt to monopolize the market by making it impossible for others to compete runs counter to the statutory purpose [of the copyright act] of promoting creative expression and cannot constitute a strong equitable basis" to disallow the fair use defense. [this is the "effect of the use upon the potential market" factor] I fully understand the desire of some companies to prevent forking of their APIs, and thus to prevent the creation of incompatible versions of programs that purport to implement those APIs. However, the copyright law is not designed to protect against such competition. Indeed, one can argue, consistent with the fundamental principles of *both* the copyright act and the open source definition, the public benefit is only served if such creative forking is allowed. I also appreciate that non-lawyers are eager to hear clear black-letter answers to questions such as the one raised originally by Bruce Behlendorf and rephrased above by Fred Patrick. I, for one, am interested in possibly seeking a black-letter decision by a judge if that will prevent companies from locking up technology through the use of the copyright law. Finally, one CAN use trademark law -- with all its strengths and weaknesses -- to prevent third parties from applying your trademark to their goods, or to prevent third parties from applying your certification mark to their incompatible goods. For example, legitimate trademarks or certification marks for "Java" or "Windows," (if they exist; I haven't searched!) could be used to prevent people from forking Java and Windows APIs and still calling those imprecise implementations "Java" or "Windows." I hope that other attorneys on this discussion list will help me evaluate whether a court challenge to the use of the copyright law to protect APIs is likely to succeed, and I hope that non-attorneys on this discussion list will help me evaluate whether obtaining that "black-letter" decision will benefit the open source movement. /Larry Rosen 650-216-1597 [EMAIL PROTECTED] www.rosenlaw.com www.opensource.org
