Jim Penny <[EMAIL PROTECTED]> writes: > On Fri, Dec 13, 2002 at 12:21:30AM +0100, Henning Makholm wrote: > > Scripsit Jim Penny <[EMAIL PROTECTED]> > > > > > So, does that not make qmail free? > > > > No. Qmail is non-free because we can't distribute modified > > *binaries*. In the case of Unicode tables, that is covered by the > > "extraction" clause. > > Let me rephrase that, is the qmail-installer not free? We can > distribute the unmodified tarball. We can (and do) use the extraction > program tar. We then mechanically patch the extracted files, build, > and install the result. > > I.e., suppose the post-inst of qmail-src invoked "build-qmail" and > "dpkg -i qmail" automatically; is that not exactly a distribute and > patch system? How is this incompatible with DJB's license? > > How does this differ from distributing UnicodeData.txt and then patching > it?
We aren't "patching UnicodeData.txt" per say, we would distribute a set of "additional characters" and "excluded characters", and programs that *read* UnicodeData.txt could read those too. Unicode permits this, since they don't make any restrictions on what *other* things a program that uses an "extraction" of unicode might do. There is, importantly, no rule that you do the extraction for the purpose of correctly implementing Unicode. In the case of qmail, there *is* a rule that prohibits just this, and so it would be a subterfuge to use an installer to achieve something that cannot be achieved directly: a binary that uses qmail source, but also isn't vanilla qmail. Unicode is *different*, because the end result of the patch (a program that uses unicode data, but also other stuff, and doesn't do Unicode at all) *is* allowed by the license. So that means that Unicode allows a program that implements a variant of Unicode, but (possibly) places a restriction on the mechanism used. Qmail does not allow the variant in the first place, and changing around the mechanism can't change that fact. Thomas
