This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [ping 3] New port: Toshiba Media Processor (mep-elf)
On Sat, 9 May 2009, DJ Delorie wrote:
> > As previously discussed I think the source for the CGEN CPU description
> > needs to be included, even if maintained elsewhere, when files generated
> > from it are included.
>
> I thought about that some more, and would like you to consider two
> points to see if they affect things:
This is ultimately a matter of FSF policy rather than a question of what I
think.
> 1. Wouldn't *all* of cgen need to be included, as the GPL requires
> that all scripts and such for building be included? CGEN is not a
> normal part of the OS or OS's development tools.
That was not the conclusion from previous discussions; we don't include
all of autoconf, or automake, or flex, or bison when we had bison parsers,
or makeinfo when distributing info files, or OCaml for the NEON intrinsics
files that have generators written in that language. We include the
sources and the generated files, but not the generator tools where those
are independent programs rather than a GCC-specific program for the
particular file being generated.
> 2. Wouldn't the requirement to include only apply to binary
> distributions, in which case people making mep-elf cross compilers
> could just include cgen in *their* source distribution? They'll
> need it to do the build anyway, since the simulator is cgen-based,
> and the mep configury tool works through cgen changes.
The point is that for GPL purposes the checked-in generated files *are*
binaries (they are not "the preferred form of the work for making
modifications to it") and so the GCC source distribution is a binary
distribution as regards these files - and so needs to include the
preferred form for modification.
--
Joseph S. Myers
joseph@codesourcery.com