Re: cygwin.dll license (was Re: FreeQt ?)
[EMAIL PROTECTED] (Mark Baker) wrote on 07.06.97 in <[EMAIL PROTECTED]>: > In article <[EMAIL PROTECTED]>, > [EMAIL PROTECTED] (Kai Henningsen) writes: > > >> Can't be linked dynamically either... read the GPL. > > > > Can too. Read the law. > > > > The GPL _cannot_ restrict someone from doing that, regardless of what they > > put in it. > > Although they _can_ restrict you from using the header files. Supposing they can (which is not quite as obvious as it looks to some people), that can of course be worked around. Most people arguing this subject seem to miss just where copyright law is coming from. Originally, this was made to protect artists - writers, painters, and so on. In these areas, borowing from other people happens fairly often. The law does not only say that some situations of borrowing need to be allowed by the original author - it also explicitely says that some need not be. It's generally a matter of degree. You've probably all heard of "fair use". In any case, unless the borrowed part is "large enough" in at least one of the work borrowed from, or the work built with it, there's nothing the original author can possibly do. I won't pretend to be able to say just how large is "large enough", except that it's a lot larger than zero. MfG Kai -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
In <[EMAIL PROTECTED]> [EMAIL PROTECTED] (Kai Henningsen) writes: > [EMAIL PROTECTED] (Jim Pick) wrote on 01.06.97 in <[EMAIL PROTECTED]>: > > > > Yes, very limiting. The code actually cannot be linked statically! > > > > Can't be linked dynamically either... read the GPL. > > Can too. Read the law. That is your opinion, the FSF's opinion is different. The FSF has shown that it is willing to defend their opinion, even if it means going to court. Unless you are also willing to go to court to defend your opinion, I think I'll side with the FSF interpretation for now. (Not that I think it's right, but I have better ways to spend my money...) -wayne -- Wayne Schlitt can not assert the truth of all statements in this article and still be consistent. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] (Kai Henningsen) writes: >> Can't be linked dynamically either... read the GPL. > > Can too. Read the law. > > The GPL _cannot_ restrict someone from doing that, regardless of what they > put in it. Although they _can_ restrict you from using the header files.
Re: cygwin.dll license (was Re: FreeQt ?)
[EMAIL PROTECTED] (Jim Pick) wrote on 01.06.97 in <[EMAIL PROTECTED]>: > > Yes, very limiting. The code actually cannot be linked statically! > > Can't be linked dynamically either... read the GPL. Can too. Read the law. The GPL _cannot_ restrict someone from doing that, regardless of what they put in it. MfG Kai -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On Jun 2, Jim Pick wrote > Just so you understand why I'm so interested - I'm working on porting dpkg > to cygwin32. Porting or re-implementing? If it's a port, dpkg is already under gpl, so cygwin32 being under gpl shouldn't be an issue. [Even if it wasn't, I don't understand how a gpl'd dll could be considered a problem.] -- Raul -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> On Jun 2, Jim Pick wrote > > Just so you understand why I'm so interested - I'm working on porting dpkg > > to cygwin32. > > Porting or re-implementing? If it's a port, dpkg is already under > gpl, so cygwin32 being under gpl shouldn't be an issue. [Even if > it wasn't, I don't understand how a gpl'd dll could be considered > a problem.] That's true. I was just thinking about all the packages that use it. It's worth doing, even if Cygnus doesn't want to LGPL their license. At least we could port the 1000+ packages in the main distribution. The non-free stuff would be questionable. Let's kill this thread - I made my point - ie. "just 'cause it's GPL'd doesn't automatically make it as 'free' as humanly possible". When I actually get dpkg to work, we can start up a new mailing list, and continue the discussion there. Cheers, - Jim pgpClFc3F4Yxq.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
Mark Eichin <[EMAIL PROTECTED]> writes: [snip] > libdb would be an issue if you used the db interfaces; if you used the > dbm_* interfaces, you'd presumably be ok... But the original libdb was covered by the BSD copyright; the libc6 copyright states: "All code incorporated from 4.4 BSD is under the following copyright: [standard BSD copyright]". db.h includes this copyright. Your point about gzip was valid, but if you're going to be picky, get it right! ;-) -- Carey Evans <*> [EMAIL PROTECTED] "Lies, damn lies, and computer documentation." -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Well, maybe the GPL is broken when it comes to situations like this. What I don't understand is, if something doesn't contain any GPL'd code, how can the GPL force me to put my product under it. So it has the interface calls to library/.dll, copyrights don't cover how something works, patents do. The GPL isn't a patent, so therefore only covers protecting the product from being reused. Well if my product doesn't explicitly use it, it shouldn't have to be under the GPL. Just my 2 cents Shaya -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Hi Jim, > Imagine if Microsoft demanded that everybody had to use a certain > license in order to run on top of their operating system. Well, they do actually. Microsoft charges for the licences to use it's ``operating systems''. If the Freeware community produces software that ends up helping closed vendors sell their wares on closed OS's, then we might end up damaging our cause. That is presumably what is behind Cygnus' attempt to put some pressure on developers to release their software as Freeware, by charging them if they don't. Unfortunately, this makes ``Debian GNU/Win32'' a rather complicated problem, to which I don't know the answer. As a Freeware bigot, I'm tempted to say ``sod them, they can't use our software unless they GPL'', but that is a probably an unenforceable (and somewhat childish) position, so what to do ? Cheers, Phil. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On 2 Jun 1997, Mark Eichin wrote: > > Now, when you link -- statically or dynamically -- you are including > > portions of libc5 in your binary. This results in your binary being > > Umm, no, actually -- the whole point of dynamic linking is that you're > *not* including portions of libc5 in your binary. A replacement libc5 > that met the "interface" of the one you used could be dropped in > instead. (#including header files, that counts -- but not linking -- > and it's sometimes surprising how much code can get away without using > the header files...) > > The same is true of .dll's and *that* is the crux of the discussion. Correct from my viewpoint > > Now that I've been informed that libc5 is really under the LGPL (or at > least parts of it claim to be) and that the /usr/doc/libc5/copyright > file is *wrong*, I can certainly see a difference between that and > cygwin32.dll. Nonetheless, neither is anything like QT. However, as far as I know, you can't statically link something a .dll under windows anyways, so it doesn't matter. The GPL is fine, and you can still use it for commercial software. > > For some more perspective on the "interface" argument, go back and see > some of the flaming a year or two ago about the GNU "libmp" (multiple > precision integer math library.) See also the discussion of just a > week or three ago about a company shipping a commercial package that > uses GNU RCS underneath -- but since GNU RCS is built as a DLL (and > they ship sources for those changes, and gnu rcs itself) they don't > have to ship the program sources (and have allegedly run this past > the FSF for confirmation that it was OK) Recall that RCS is > GPLed, not LGPLed. > > Isn't this fascinating? :-) I must admit that I'm glad to see, all in > all, that this discussion has stayed *so* polite in comparison to the > typical gnu.misc.discuss or other open net thread. Thanks! > Me too. Shaya -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On Jun 2, Raul Miller wrote > > [Note: what RMS is trying to argue against is the stunt > Steve Jobs & Co. pulled with Objective C.] Could you describe what the said 'stunt' was? I'm curious... Christian pgpyv2Q82qumI.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
> On Jun 2, Jim Pick wrote > > The cygwin.dll case in an example where the GPL is being used to restrict > > the > > rights of other people using the code so that they can't do something taboo > > such as charge money, while at the same time, reserving the right for the > > authors to do the exact same thing. To me, this is clearly hypocritical, > > and I don't consider that software to be as 'free' as it could otherwise > > be. > > First off, this list isn't the right forum to discuss Cygnus morality > issues. Can someone point out a better forum? I'm not saying that they're being immoral. I don't think they have properly addressed the issues though. Maybe that means they would be open to releasing the cygwin.dll under the LGPL in addition to the GPL and their proprietary license. > Second, I find it hard to conceive of some case wher Cygnus would > sue someone for selling commercial software which happened to use > a DLL authored by Cygnus. It would trash their (Cygnus's) reputation, > and eat into their bottom line. Cygnus has made it clear that they intend to make money off of cygwin32. How aggressively they do that, I don't know. > Third, I think you're (Jim, I mean) making a mountain out of a mole hill. Perhaps. Cygnus hasn't released enough information for me to decide whether it is a mountain or a mole hill. I hope it's a mole hill. Just so you understand why I'm so interested - I'm working on porting dpkg to cygwin32. That way, we'll be able to host the entire Debian distribution on top of Windows 95 and Windows NT (at least the stuff that will port). It would just be another Debian port, like PowerPC, Sparc or Alpha. This could potentially be a really big thing. :-) Little licensing details could really come back to haunt us. Imagine if everybody that wanted to make a non-free application that ran on top of Debian GNU/Win32 had to pay Cygnus a licensing fee. Imagine if Microsoft demanded that everybody had to use a certain license in order to run on top of their operating system. > Can't we talk about something more interesting? This is interesting! :-) (Nobody's forcing you to read this thread) Cheers, - Jim pgpv1yZd6vYvT.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
On Jun 2, Jim Pick wrote > The cygwin.dll case in an example where the GPL is being used to restrict the > rights of other people using the code so that they can't do something taboo > such as charge money, while at the same time, reserving the right for the > authors to do the exact same thing. To me, this is clearly hypocritical, > and I don't consider that software to be as 'free' as it could otherwise > be. First off, this list isn't the right forum to discuss Cygnus morality issues. Can someone point out a better forum? Second, I find it hard to conceive of some case wher Cygnus would sue someone for selling commercial software which happened to use a DLL authored by Cygnus. It would trash their (Cygnus's) reputation, and eat into their bottom line. Third, I think you're (Jim, I mean) making a mountain out of a mole hill. Can't we talk about something more interesting? Like, a mechanism for informing maintainers of packages what issues they need to address to get packages out of Incoming and into the distribution? -- Raul -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Jason Gunthorpe wrote: > I really must admit I find the GPL very cryptic, it's hard to say exactly > what it means if you look at very small detail. I do think that it makes > sense however that you should be able to put RCS in a dll and link to the > dll. That depends, if you put it in a .dll, and the original author is just a student, or a hobbiest, it's unlikely that you would ever have to prove your point in court. But if the original author is a commercial entity that is trying to make money, or perhaps the FSF that has a point to prove, you might find yourself in court, with a bunch of expensive lawyers on the other side. Cheers, - Jim pgpdA77lNmXjz.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
> [ I've not been following this thread too closely, > so if I've got the wrong idea, please forgive me ] > > > The GPL is a very restrictive license. In many ways, it is just as > > restrictive as the Qt license. Particularily in the case of libraries, > > using it as Cygnus is doing (to make money) goes against the spirit > > of Free Software. > > Wrong. (I think I'm right) > There is no obligation to give things away for no money when writing free > software. No, there isn't an obligation. There isn't an obligation to even have to write free software. I have no problem with people who write proprietary software -- something's got to pay the bills. But there are varying degrees of freedom. There exists "Free Software" where somebody isn't trying to make a buck off of it. Most "Free Software" falls into this category. The GPL license is used by many of these packages in order to prevent anybody from putting the software under a proprietary license in order to 'extort' money (and other things) from out of the user base. The cygwin.dll case in an example where the GPL is being used to restrict the rights of other people using the code so that they can't do something taboo such as charge money, while at the same time, reserving the right for the authors to do the exact same thing. To me, this is clearly hypocritical, and I don't consider that software to be as 'free' as it could otherwise be. If cygwin.dll was put under the LGPL, it would be a more 'free' piece of software that if it was under the GPL. But then Cygnus couldn't 'extort' money from their users (some of whom may be writing commercial software to put food on the table for their kids). [I use the word, 'extort' in a Free Software sense, since the library is being passed off as Free Software] There's something wrong with thinking that just because something is under the GPL, it is automatically as 'free' as is could be. > I presume that the what they are selling is the right not to be bound by the > GPL restrictions that would normally apply --- is that correct ? That's true. But if there is a great demand for relaxed restrictions, a true-blue free software author would investigate using a less restrictive license, such as the LGPL, rather than prying money out of the hands of the users. (hopefully I'm clearing up some people's thinking on this topic) Cheers, - Jim pgpl9QeB0Kulz.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
> However, the unique interface issue does exist with regard to gzip, > since that is purely a GPLed product. I think a libgzip or a gzip.dll > would run into the same issues as the libdb did. Not to distract from the original point (thank you for the clearer explanation of the libmp issue!) note that "zlib", which uses the same algorithm, is an unencumbered implementation (more suited for embedding anyway, which makes a gzip.dll simply a poor choice :-) and thus the whole issue is fairly well side stepped. (X is using zlib for both low-bandwidth-X and for font compression now...) libdb would be an issue if you used the db interfaces; if you used the dbm_* interfaces, you'd presumably be ok... -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Buddha Buck wrote: >However, the unique interface issue does exist with regard to gzip, >since that is purely a GPLed product. I think a libgzip or a gzip.dll >would run into the same issues as the libdb did. The source code to the zlib library has been released together with ssh with a non-GPL license (pretty much BSD-like). -- Thomas Koenig, [EMAIL PROTECTED], [EMAIL PROTECTED] The joy of engineering is to find a straight line on a double logarithmic diagram. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> > On 2 Jun 1997, Mark Eichin wrote: > > > For some more perspective on the "interface" argument, go back and see > > some of the flaming a year or two ago about the GNU "libmp" (multiple > > precision integer math library.) See also the discussion of just a > > week or three ago about a company shipping a commercial package that > > uses GNU RCS underneath -- but since GNU RCS is built as a DLL (and > > they ship sources for those changes, and gnu rcs itself) they don't > > have to ship the program sources (and have allegedly run this past > > the FSF for confirmation that it was OK) Recall that RCS is > > GPLed, not LGPLed. > > Hm, that's very interesting. Someone I was talking with a time back used > the example 'Putting GZIP in a dll and then linking to it still makes your > code GPL'. But if the FSF says that it is okay to do that then it is okay > to do that ;> I'm not familiar with the RCS debate, but I was reading gnu.misc.discuss during the libmp situation. Based on that debate, I can see why rcs.dll might be allowed, but gzip.dll might not. The issue in the libmp was a package containing a midified RSAREF that could be linked to libmp. Libmp is aparantly faster than the standard multiprecision library available. Libmp also has a slightly different interface, so it isn't a simple drop-in replacement for the standard library (as glibc or libc5 (theoretically) is). The FSF contended that the resulting modified package (which was not distributed with binaries or source for libmp) must be GPLed, since the -product-, namely the executable binaries, must contain GPLed code (the libmp library), so must be GPLed. The source is merely the preferred distribution method for the product. In this case, the product was being distributed in two pieces. The justification for this position was that libmp had a unique interface. Any program written to use that interface had no choice but to use libmp, and thus the resulting binary was derived from libmp. In this particular case, the program was thus subjected to both the GPL -and- the license on RSAREF, which are incompatable licenses. The FSF objected to the distribution of the modified package -at all-, since it would be impossible to fulfill the requirements of both licenses. That particular package is now distributed with a simple libmp-compatable non-GPLed multiprecision integer package (thus avoiding the unique interface issue, since now there are two libraries with the same documented interface), and instructions to link it with the FSF libmp, because it is a much better library. RMS agreed that this would solve the problem. Applying that to rcs.dll, it seems to me that as long as the dll doesn't rely on any GNU-specific RCS feature, then it would be providing a non-unique, standard interface. Two dll's could exist -- one based on GNU rcs, and the other that makes the appropriate system() calls (or whatever the Windows equivilant is) to do the job. If the latter is in fact what the dll does, requiring separate installation of an appropriate RCS package, then it obviously doesn't have the same encumberance problems as the libmp did. However, the unique interface issue does exist with regard to gzip, since that is purely a GPLed product. I think a libgzip or a gzip.dll would run into the same issues as the libdb did. > I really must admit I find the GPL very cryptic, it's hard to say exactly > what it means if you look at very small detail. I do think that it makes > sense however that you should be able to put RCS in a dll and link to the > dll. The debate around that is all based on the question of what is a > derived work. One could even argue executing gzip in a pipeline makes > other elements in the pipeline 'derived' somehow from gzip. The GPL just > doesn't make that perfectly clear! There are a lot of unclear issues, unfortunately. I think that there are at least 4 different issues here: 1) what the FSF and RMS want, 2) what their lawyers think they can get away with using the license, 3) reasonable lay interpretations of the license, and 4) judicial interpretation of the license. The second point implies subterfuge on the part of the lawyers or RMS. I don't think so. I think RMS has made it perfectly clear what he wants: a complete overhaul of the intellectual property system with regards to software in the vain hope of returning to the free and open early days of the labs at MIT. But his lawyers must work -within- the existing IP system to subvert it. They believe (and are staking their professional reputation on it) that the GPL represents the closest approximation of RMS's desires (of a complete subversion of IP law) within the framework of existing law. It is always tricky to subvert a structure from within, and that is why the GPL is so tricky to interpret. However, it is item 4) that is the key, and the GPL has (to my knowledge) never been tested in court. Perh
Re: cygwin.dll license (was Re: FreeQt ?)
[EMAIL PROTECTED] (Jason Gunthorpe) wrote on 01.06.97 in <[EMAIL PROTECTED]>: > I really must admit I find the GPL very cryptic, it's hard to say exactly > what it means if you look at very small detail. I do think that it makes > sense however that you should be able to put RCS in a dll and link to the > dll. The debate around that is all based on the question of what is a > derived work. One could even argue executing gzip in a pipeline makes > other elements in the pipeline 'derived' somehow from gzip. The GPL just > doesn't make that perfectly clear! Of course, it's actually not the job of the (L)GPL to define "derived work", and all experts I've heard seem to agree that they made a botch of it. The term is defined by law (and international treaty), and it seems quite clear that putting parts from one work into another, where these parts are small with respect to both the first and the second work, definitely DO NOT make the second one a derived work, whatever any license may claim. Think about where this comes from. If I write a book, and include Hamlet's famous question somewhere, my book is not a derived work from Shakespeare's. Now, you can of course argue about how large some peaces are - if I put half of Hamlet in my book, and this makes out half of my book, then it certainly _is_ a derived work. But nothing Shakespeare could have said about derived works (assuming he wasn't dead long before this term was invented) can possibly change that. MfG Kai -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On Jun 1, Jim Pick wrote > Actually, I had a very similar polite argument with RMS via private e-mail > (about linking Java libs with mixed GPL/LGPL/proprietary licenses). He > was pretty solid on the fact that run-time linking is the same as > "compiled-in" linking. Yep, once the run-time linking has occured you're not allowed to redistribute the resulting image if you aren't willing to redistribute the source under similar terms. This isn't that big of an issue for most people. [Note: what RMS is trying to argue against is the stunt Steve Jobs & Co. pulled with Objective C.] -- Raul -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
[ I've not been following this thread too closely, so if I've got the wrong idea, please forgive me ] > The GPL is a very restrictive license. In many ways, it is just as > restrictive as the Qt license. Particularily in the case of libraries, > using it as Cygnus is doing (to make money) goes against the spirit > of Free Software. Wrong. There is no obligation to give things away for no money when writing free software. The word ``free'' here applies to the free-ly available source, which you are allowed to take, and modify, and maintain yourself if you wish, and you can then sell it for lots of money, as long as the people you sell it to also get the source, and the right to modify, maintain and sell it, with the proviso, etc. etc. The main evil that RMS was trying to combat with GPL was the fact that people regularly get left with software for which they do not have the source, and find that they can not get support from the original supplier for one reason or another (gone bust, moved on to new versions etc.). I suppose the thing that Cygnus seem to have done that might be morally wrong is to take patches written in the freeware spirit, and started selling them because they hold the copyright to the work as a whole. I presume that the what they are selling is the right not to be bound by the GPL restrictions that would normally apply --- is that correct ? If they are actually maintaining two source trees, and stealing ideas from the GPL source to enhance the commercial version, then I think they are in the wrong, but I cannot imagine they would be doing that. Cheers, Phil. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> For some more perspective on the "interface" argument, go back and see > some of the flaming a year or two ago about the GNU "libmp" (multiple > precision integer math library.) Actually, I had a very similar polite argument with RMS via private e-mail (about linking Java libs with mixed GPL/LGPL/proprietary licenses). He was pretty solid on the fact that run-time linking is the same as "compiled-in" linking. What I think it comes down to is this -- if the GPL'd code comes from a company that is willing to hire lawyers -- you'd better pay attention to the fine print, otherwise, don't worry about it that much. I'm sure that there are plenty of libraries out there that have been put under the GPL, because the author couldn't be bothered to worry about the implications. I've seen a few Java ones that fit this bill. You could probably use these in a commercial app, and nobody would care. The Linux kernel is GPL'd, but proprietary stuff gets dynamically linked to it indirectly via OS calls and such. This hasn't been an issue, since Linus Torvalds isn't going to sue you. The FreeBSD guys would have you believing otherwise. Cygnus is trying to sell commercial licenses, so that implies that they would be willing to sue. This is going to be an issue for us, the Debian project, when I finish porting dpkg to cygwin32. The GPL was a quick hack designed to cover stand-alone apps. It was never intended to be used for libraries and other dynamically-linked code where the legal implications are much more far-reaching. That's why the LGPL came into existence - the GPL was just too restrictive. The GPL is a very restrictive license. In many ways, it is just as restrictive as the Qt license. Particularily in the case of libraries, using it as Cygnus is doing (to make money) goes against the spirit of Free Software. At least with Qt, Troll Tech is very up-front about the fact that it is commercial software, which they are licensing for free. Cygnus, on the other hand, called their work the "GNU-Win32 project", promoted it as genuine true-blue GPL'd "Free Software", solicited patches from the user community, and then, after 17 betas or so (maybe not all public), they issued a marketing announcement that "commercial" licenses could be arranged. Many people on the mailing list were not impressed -- they felt that they had been cheated. Don't get me wrong, I like the work Geoffrey Noer and others have done -- I'm still going to use it. But I don't consider it to be "Free Software" in spirit, even if it is under the GPL. I'd like to see Debian maintain some lofty goals as to what constitutes "Free Software", so I think that discussion on these topics is healthy. Just calling 'em like I see 'em. Cheers, - Jim pgp2R1wJKPNJd.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
On 2 Jun 1997, Mark Eichin wrote: > For some more perspective on the "interface" argument, go back and see > some of the flaming a year or two ago about the GNU "libmp" (multiple > precision integer math library.) See also the discussion of just a > week or three ago about a company shipping a commercial package that > uses GNU RCS underneath -- but since GNU RCS is built as a DLL (and > they ship sources for those changes, and gnu rcs itself) they don't > have to ship the program sources (and have allegedly run this past > the FSF for confirmation that it was OK) Recall that RCS is > GPLed, not LGPLed. Hm, that's very interesting. Someone I was talking with a time back used the example 'Putting GZIP in a dll and then linking to it still makes your code GPL'. But if the FSF says that it is okay to do that then it is okay to do that ;> The other neat GPL issue comes in with C++, you actually DO include instances of code in your program with inlines, templates, vtables and other things. Fortunately G++ is completely free if compiled and used with GNU's compiler, LGPL otherwise. I really must admit I find the GPL very cryptic, it's hard to say exactly what it means if you look at very small detail. I do think that it makes sense however that you should be able to put RCS in a dll and link to the dll. The debate around that is all based on the question of what is a derived work. One could even argue executing gzip in a pipeline makes other elements in the pipeline 'derived' somehow from gzip. The GPL just doesn't make that perfectly clear! Jason -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> Now, when you link -- statically or dynamically -- you are including > portions of libc5 in your binary. This results in your binary being Umm, no, actually -- the whole point of dynamic linking is that you're *not* including portions of libc5 in your binary. A replacement libc5 that met the "interface" of the one you used could be dropped in instead. (#including header files, that counts -- but not linking -- and it's sometimes surprising how much code can get away without using the header files...) The same is true of .dll's and *that* is the crux of the discussion. Now that I've been informed that libc5 is really under the LGPL (or at least parts of it claim to be) and that the /usr/doc/libc5/copyright file is *wrong*, I can certainly see a difference between that and cygwin32.dll. Nonetheless, neither is anything like QT. For some more perspective on the "interface" argument, go back and see some of the flaming a year or two ago about the GNU "libmp" (multiple precision integer math library.) See also the discussion of just a week or three ago about a company shipping a commercial package that uses GNU RCS underneath -- but since GNU RCS is built as a DLL (and they ship sources for those changes, and gnu rcs itself) they don't have to ship the program sources (and have allegedly run this past the FSF for confirmation that it was OK) Recall that RCS is GPLed, not LGPLed. Isn't this fascinating? :-) I must admit that I'm glad to see, all in all, that this discussion has stayed *so* polite in comparison to the typical gnu.misc.discuss or other open net thread. Thanks! -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On 2 Jun 1997, Kai Henningsen wrote: > [EMAIL PROTECTED] (Jason Gunthorpe) wrote on 01.06.97 in <[EMAIL PROTECTED]>: > > > On 1 Jun 1997, Mark Eichin wrote: > > > > > > I believe libc5.so is LGPL... > > > > > > I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at > > > all*, though the libc6 one mentions both. > > > > Yep, the copyright file does not mention the LGPL at all. This seems to me > > to be very limiting of commercial software running on linux. > > Yes, very limiting. The code actually cannot be linked statically! What a > tragedy ... NOT. There seems to be some confusion here. The GPL states that when GPL code is aggregated with non GPL code the new code is covered by the GPL when they are combined (what this means I am still unsure), ie: --- These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based ^^^ This bit on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. --- Now, when you link -- statically or dynamically -- you are including portions of libc5 in your binary. This results in your binary being covered under the GPL. I am not sure how that will effect the source code. The common belief is that it forces the source code to be included (though likely not GPL'd) with the binary. If you use a LGPL'd library then statically linking requires that you destribute relinkable object form versions of your binary so the user can upgrade the statically linked lib. Jason -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On Sun, 1 Jun 1997, Galen Hazelwood wrote: > Jason Gunthorpe wrote: > > > > On 1 Jun 1997, Mark Eichin wrote: > > > > > > I believe libc5.so is LGPL... > > > > > > I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at > > > all*, though the libc6 one mentions both. > > > > Yep, the copyright file does not mention the LGPL at all. This seems to me > > to be very limiting of commercial software running on linux. > > I believe that regardless of what our copyright file says, glibc 1.0 > (libc5) and 2.0 (libc6) are both LGPL--at least the library parts. > Other programs grouped with the libc package are probably GPL. Ack! I must be blind, I looked right at this file right before posting too, from stdio.h: This library is free software; you can redistribute it and/or modify it under the terms of the GNU Library General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. Right there, 2nd line 'GNU Library General'. /usr/doc/copyright/libc5 says GPL not LGPL. Sounds like a bug in the libc5 package!! Jason -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> I just brought this up, since it was my understanding that if you > want to write a commercial program (ie. not under the GPL), and > link it against cygwin.dll, you've got to pay Cygnus $$$. Not all > that different than the restrictions on Qt, really. Actually, it is different. GPL-ed software gives you the right to change the source, and gives you right to link other GPL-ed software to it on all platforms. Debian doesn't presently have a rule against libraries that pass the GPL infection, although we prefer to avoid them. Thanks Bruce -- Bruce Perens K6BP [EMAIL PROTECTED] 510-215-3502 Finger [EMAIL PROTECTED] for PGP public key. PGP fingerprint = 88 6A 15 D0 65 D4 A3 A6 1F 89 6A 76 95 24 87 B3 -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
On Sun, 1 Jun 1997, Jim Pick wrote: > > > Yes, very limiting. The code actually cannot be linked statically! > > Can't be linked dynamically either... read the GPL. > I'm not sure from a copyright standpoint how that works. A copyright means that you are protected from me using your copyrighted item. Well, if I don't give libc or any other gpl'd library away, be it as a statically linked app, or by giving away the shared library how am I violating the gpl. If joe end-user already has the library, how am I violating the copyright. Even if commercial products build against a gpl'd library, if they are only linked dynamically against the library, i.e. they don't contain any code from the library, and that library can be replaced by another one (look at lesstif vs. Motif). In my view LPGL=I can statically link my applications to the library and sell it w/o source code. GPL=I can statically link my application to the library, but my application now has to be GPL'd because it contains GPL'd code. However, if it is only dynamically linked, since it doesn't contain any GPL'd code, I can sell it as a commercial app w/o giving out source code. Shaya -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> Yes, very limiting. The code actually cannot be linked statically! Can't be linked dynamically either... read the GPL. Cheers, - Jim pgp6b75kk1gUm.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
[EMAIL PROTECTED] (Jason Gunthorpe) wrote on 01.06.97 in <[EMAIL PROTECTED]>: > On 1 Jun 1997, Mark Eichin wrote: > > > > I believe libc5.so is LGPL... > > > > I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at > > all*, though the libc6 one mentions both. > > Yep, the copyright file does not mention the LGPL at all. This seems to me > to be very limiting of commercial software running on linux. Yes, very limiting. The code actually cannot be linked statically! What a tragedy ... NOT. MfG Kai -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Jason Gunthorpe wrote: > > On 1 Jun 1997, Mark Eichin wrote: > > > > I believe libc5.so is LGPL... > > > > I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at > > all*, though the libc6 one mentions both. > > Yep, the copyright file does not mention the LGPL at all. This seems to me > to be very limiting of commercial software running on linux. I believe that regardless of what our copyright file says, glibc 1.0 (libc5) and 2.0 (libc6) are both LGPL--at least the library parts. Other programs grouped with the libc package are probably GPL. If our copyright file says otherwise, our copyright file is wrong. This should be looked into. I'd grab the source and check myself, but it takes a long time over a 28.8k line. --Galen -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> Two questions: (1) in what way is cygwin32.dll different from libc5.so > in this regard (since the license for both is the same: GPL) libc5 appears to be under the GPL, while libc6 appears to be under the LGPL. Weird. Does that mean that anything that is linked against libc5 has to be GPL'd? I'm surprised I haven't heard more about this - I obviously don't know the whole story. Maybe there is just widespread abuse of the GPL. If everyone is just ignoring it, that doesn't provide much legal protection for Cygnus if they're trying to make money off of cygwin.dll. > (2) the discussion wasn't writing *comercial* software with > anything, but writing *free* software with a pseudo-free package like > Qt... so how did we get here? There's *certainly* no problem writing > gpl'ed software with cygwin32.dll :-) There's not really any problem writing *free* software with Qt either. That's why I deliberately confused them together... Free software shouldn't be about confusion. > ps. A friend of mine with whom I've been discussing this says that > if we took all the time we've spent flaming about this and actually > *wrote some code* we wouldn't have the problem in the first place :-) I am working with cygwin.dll right now actually, trying to get dpkg to port to it (well, trying to hack it so I can get Perl to go, so then I can attempt to build dpkg). Klee is going to update the cross-compiler to assist me. Hopefully, cygwin.dll can become a part of the Debian distribution for a Win32 port, playing the same role as the Linux kernel. But it would be a shame if we have to reclassify the copyrights on every package in the distribution (and prohibit non-free stuff) just because of it. Cheers, - Jim pgpyPc6e7CsBE.pgp Description: PGP signature
Re: cygwin.dll license (was Re: FreeQt ?)
On 1 Jun 1997, Mark Eichin wrote: > > I believe libc5.so is LGPL... > > I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at > all*, though the libc6 one mentions both. Yep, the copyright file does not mention the LGPL at all. This seems to me to be very limiting of commercial software running on linux. >From the GPL section 2: These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. I don't know what to make of that, it sounds like the binary has to be 'free'? Which would in turn mean all debian programs are 'free'? Bah, I have got to quit reading this GPL! Jason -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> I believe libc5.so is LGPL... I don't. /usr/doc/libc5//copyright doesn't *mention* the LGPL *at all*, though the libc6 one mentions both. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
Mark Eichin wrote: > > > I just brought this up, since it was my understanding that if you > > want to write a commercial program (ie. not under the GPL), and > > link it against cygwin.dll, you've got to pay Cygnus $$$. Not all > > that different than the restrictions on Qt, really. > > Two questions: (1) in what way is cygwin32.dll different from libc5.so > in this regard (since the license for both is the same: GPL) I believe libc5.so is LGPL... --Galen -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: cygwin.dll license (was Re: FreeQt ?)
> I just brought this up, since it was my understanding that if you > want to write a commercial program (ie. not under the GPL), and > link it against cygwin.dll, you've got to pay Cygnus $$$. Not all > that different than the restrictions on Qt, really. Two questions: (1) in what way is cygwin32.dll different from libc5.so in this regard (since the license for both is the same: GPL) (2) the discussion wasn't writing *comercial* software with anything, but writing *free* software with a pseudo-free package like Qt... so how did we get here? There's *certainly* no problem writing gpl'ed software with cygwin32.dll :-) [I'm not representing Cygnus in this; though I've used and hacked on cygwin32, all of my current Cygnus work [Kerberos in particular] is under an X11-style license, though Federal Regulations make it "difficult" to redistribute...] ps. A friend of mine with whom I've been discussing this says that if we took all the time we've spent flaming about this and actually *wrote some code* we wouldn't have the problem in the first place :-) -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
cygwin.dll license (was Re: FreeQt ?)
> yeah, cygwin32.dll is under the GPL. So? It's a DLL, like libc5 and > libc6 are... [the *only* thing I'm aware of that actually uses the > LGPL is libg++; it was as much of an experiment as anything, and I'm > not aware of any not-otherwise-free software taking advantage of those > terms...] Just because libgcc is on "special" terms is no reason for > cygwin32.dll to be (cygwin32 is *more* than even a libc, it's got a > fair amount of emulation code in it, so it looks like you have unified > file descriptors... and you don't want to look at the internals of > select...] I just brought this up, since it was my understanding that if you want to write a commercial program (ie. not under the GPL), and link it against cygwin.dll, you've got to pay Cygnus $$$. Not all that different than the restrictions on Qt, really. I don't think this situation exists with libc5 or libc6 (ie. Netscape and Sun's JDK are linked against it). I'm not familiar with the licenses on everything though -- I hate reading the fine print. If I'm wrong on this issue (I hope I am), please correct me. Cheers, - Jim pgpLjYTjNzXwK.pgp Description: PGP signature