Alexander Cherepanov <[EMAIL PROTECTED]> wrote: > 15-Dec-03 07:39 Walter Landry wrote: > > Alexander Cherepanov <[EMAIL PROTECTED]> wrote: > >> 8-Dec-03 20:43 Walter Landry wrote: > >> Thus, when distributing binaries compiled from sources, the > >> compilation is under Section 2 and the distribution is under Section > >> 3. That's part of the original confusion--requirement of source form > >> in Section 2 applies only to distribution, not to modification. > > > Compilation is not covered by the license, only distribution. > > Distribution of modified binaries is covered by Sections 2 and 3. > > Section 3 lets distribute binaries as long as you distribute source. > > Section 2 tells you what you have to do if you modified the source. > > Pardon? You agreed above that compilation is a kind of modification. > And modification is clearly covered by Section 2. > Or you mean that Section 3 implicitly contains a permission to > compile? Or what?
I mean that Section 3 lets you distribute binaries as long as you distribute source. If you happen to have modified that source, then you have to comply with Section 2 in how you've modified the source (prominent notices etc.) > >> > Distributing binaries under Section 2 probably means > >> > editting the binaries with a hex editor. You also need to have the > >> > rights to distribute everything in the binary under the GPL. > > BTW, that's also interesting. Can you distribute, under Section 3, a > binary statically linked with a library from non-free compiler? > Probably not. Section 3 clearly says that the binary must be > distributed under the terms Sections 1 and 2. And 2b says that the > whole thing must be under the GPL. If the library linked in is, for > example, non-modifiable that's impossible. Though one can argue that > this is the usual way of creating an executable and therefore Section > 3 gives a permission to do just that... Section 3 specifically lists the compiler as one of the things that you don't have to distribute under the GPL. Historically, many things were compiled with a non-free compiler and distributed by the FSF under the GPL. > >> > With > >> > non-free compilers, that may be a problem. With gcc, that probably > >> > means more hex editing to include the FSF, HP, SGI, etc. copyrights. > >> > >> The only difference in distribution under Section 2 and under Section > >> 3 is the requirement for sources. > > > True. That would seem to imply that you have to preserve copyright > > notices etc. in all of the modified files. Otherwise it makes no > > sense to refer to the terms of Section 1. > > If it makes no sense to refer to it from Section 2 then it makes no > sense to refer to it from Section 3 at all. > > In fact, Section 1 contains the following requirements: > > (a) you conspicuously and appropriately publish on each copy an > appropriate copyright notice and disclaimer of warranty; > > (b) keep intact all the notices that refer to this License and to > the absence of any warranty; and > > (c) give any other recipients of the Program a copy of this License > along with the Program. > > You seems to talk only about (b) but (a) and (c) are quite reasonable > in any situation (with "the Program" in (c) replaced by something > else, of course). Exactly. But those are much easier to comply with. > > But the usual way of > > creating an executable is by running code through a compiler, which > > removes most of those notices. So you could argue that Section 3 > > gives you permission to do just that. > > Now I get your idea. You say that one usually cannot distribute a > binary under Section 2 because he cannot comply with some technical > requirements from Section 1 but these requirements are overridden when > distributing under Section 3. That would be an interesting solution. Yup. > >> The hole in the explicit wording seems to be so clear that I start > >> doubting it is just an oversight. Maybe it's normal for sections of a > >> license to trump each other? > > > The hole is there, but exploiting it is hard. People don't normally > > modify machine code. > > That's true for machine code but there are other kinds of binaries. > Consider info for example. It usually contains copyright notices and > it usually is not source. Is it ok to distribute .info without .texi? As long as it complies with Section 2, it should be fine. I don't know if, in general, .info files will comply. For example, does any copyright from makeinfo leak into the .info files? Regards, Walter Landry [EMAIL PROTECTED]
