Date: Mon, 25 Mar 2002 13:59:22 +0300 From: Hans Reiser <reiser@namesys.com> To: Terry Lambert <tlambert2@mindspring.com> Cc: hiten@uk.FreeBSD.org, Chris Mason <mason@suse.com>, Josh MacDonald <jmacd@CS.Berkeley.EDU>, Parity Error <bootup@mail.ru>, freebsd-fs@FreeBSD.ORG, reiserfs-dev@namesys.com Subject: Re: [reiserfs-dev] Re: metadata update durability ordering/soft updates Message-ID: <3C9F030A.7040205@namesys.com> References: <20020317225759.82774.qmail@web21109.mail.yahoo.com> <3C95ACBA.4040108@namesys.com> <3C95B838.F8ECE39A@mindspring.com> <3C95C8C3.7080803@namesys.com> <3C966CDF.25A7A379@mindspring.com> <3C9E1D6E.3080604@namesys.com> <3C9E6BEC.B2EB8D86@mindspring.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Terry Lambert wrote: >Hans Reiser wrote: > >>Terry Lambert wrote: >> >Precisely. So you can port it, and the result of the port >> >is still GPL'ed. At that point, you can treat it like any >> >other GPL'ed code that the original vendor had ported. >> > >> >You do the port, buy you don't distribute it, so you are not >> >required to make sources available. You merely use the port >> >internally. >> > >> >Alternately, you do the port, you distribute it, but you do >> >>distribute means what? >> > > "to give out or deliver especially to members of a > group <distribute newspapers> <distribute leaflets>" > >If you never do this, then the clauses which require source to >be given out are never invoked. > >> >not distribute it linked against your proprietary code. You >> >make the end user do the linking, if they want to use it. By >> >>There is absolutely nothing in the license that makes derivative >>specific to linking issues. >> > >It's not an issue of derivation. It's an issue of linking. It's >put in plain English in the last paragraph: > > This General Public License does not permit incorporating > your program into proprietary programs. If your program > is a subroutine library, you may consider it more useful > to permit linking proprietary applications with the > library. If this is what you want to do, use the GNU > Library General Public License instead of this License. > >If I link A + B -> C, then C is a derivative work of A and C is >a derivative work of B. > >If you look at clause 2(b), you'll see: > > You must cause any work that you distribute or publish, > that in whole or in part contains or is derived from the > Program or any part thereof, to be licensed as a whole at > no charge to all third parties under the terms of this > License. > >The "in whole or in part contains" is the kicker for linking. > To contain does not require linking. > > > >> >optioning it, you are off the hook for making your proprietary >> >source code available, but the end user is not. Since the >> >end user never had your source code, the end user can not >> >distribute the combined code further. >> >>No, you have made a derivative work. That you linked it only through a >>surrogate means nothing. >> > >The user is not a surrogate. If you had written a program, and >that did the linkage, then the program would be a surrogate. By >leaving it as an exercise for the user, you are in the same >position as an distillery or gun or a bolt cutter manufacturer: >not responsible for the use to which tsomeone puts your product. > > >>Consider what will happen as we move to NUMA/cluster/distributed >>computing architectures. Linking will become less and less meaningful, >>because programs will be composed into wholes much as functions are >>composed into wholes currently. This is perhaps why the GPL doesn't >>say that if it isn't linked, it is okay. >> > >At this point, they will have to use the "escape to a new version" >mechanism in clause 9 to keep the license relevent. The initial >LGPL failed to take into account dynamic linking; it still doesn't >take into account data interfaces, or initialized data, which end >up as part of the program image (the curses library has a number >of external variables, and there are always declarations changes on >"errno" and "syserrlist[]"). > > >>I think it is a real problem that people can't easily get a good >>authoritative definition of what is and is not a derivative work >>when they need to make their decisions. >> > >At least in the U.S., there is a very clear legal definition, >having to do with both overall percentage of code, and centrality >of the code to the resulting work. > >-- Terry > > So, if a vendor (not being hypothetical here) bases a cluster filesystem on reiserfs, is reiserfs central to the resulting work? I think so. I also think that linking or not linking is not determinative. Hans To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C9F030A.7040205>