Bug#27906: PROPOSED] Binary-only NMU's

[Santiago Vila <sanvila@unex.es>]

Ian,

before you propose a complete reorganization of our FTP archive to
"comply" with the GPL, please take a look at the "SOURCES" file in the GNU
operating system, version 0.2.

Some excerpts:

*-------------------------------------------------------------------
Sources for binaries in GNU version 0.2.

Typical configure line is

../../src/foo-NN/configure --prefix= --cache-file=../config.cache i486-gnu

[...]

cvs (1.9)
  [ inhibit use of libc getopt by defining _LIBC in lib/getopt.c
    and lib/getopt1.c immediately before it is tested. ]

[...]

findutils (4.1)
  [ Comment out basename in find/defs.h and find/util.c.
    Define _GNU_SOURCE at front of find/pred.c.
    Define ARG_MAX to be 20480 before its first use. ]

[...]

gzip (1.2.4)
  [ Comment out basename in gzip.h and util.c. ]

[...]

*---------------------------------------------------------------------

GPL says:

The source code for a work means the preferred form of the work for
making modifications to it.  For an executable work, complete source
code means all the source code for all modules it contains, plus any
associated interface definition files, plus the scripts used to
control compilation and installation of the executable.


Ok, what happens if the compilation is not done entirely by scripts, but
with a combination of scripts + manual hacking, as described by this file?

Do we always have to compile something by scripts?

[ Today I have uploaded four packages for hurd-i386, in all the four cases
I have hardcoded the Depends line by hand, because we have not a working
cross-dpkg-shlibdeps among our cross-compiler tools yet. The day we have
one, the end result will be the same. Would I be violating some license if
the package were GPLed just for not providing a patch somewhere? ].


Is the FSF violating its own license by "describing with words" how to
compile the various packages in the GNU system?

-- 
 "72af48bcad139b59857329e82270e80a" (a truly random sig)