Is this better for tomsrtbt?
tom at toms.net
Sat Apr 21 18:37:24 UTC 2001
> What is it you want to protect?
I'll give you an example.
My bootdisk currently includes a libc5.so.5.4.13 that I have down to only
416,361 bytes. More than 2 years ago, back in October of 1998, I had it
only down to 432,684 bytes. I havn't distributed that binary libc.so for
about 2 years. Now, guess what libc5.so.5.4.13 file is currently in use
on the MuLinux distribution? Yep, the 432,684 byte one I created about
2 years ago. Now, the author of MuLinux does *not* mention that he used
a bunch (more than just this) of stuff directly out of tomsrtbt, he did
*not* ever contact me about how to build it that small, and I would be
curious in the abstract to know what he would do if someone asked him
for the source code sufficient to rebuild it. Now, it is within 3 years
of when I distributed it- so he could still ask me for the source to the
libc I distributed 2 years ago- and he does mention tomsrtbt positively
as another minilinux to check out- but I think it is fair that I require
that users of anything I *can* require this of, give me credit. (Note:
this is not meant to single out MuLinux, there are others doing the same
thing). At a MINIMUM, it is only fair that a user of one of these know
where the stuff came from, after all, what if I stuck a trojan in it? I
really want any object file created by a compiler on my machine to be so
documented. It isn't fair that users of these distributions are allowed
to have the impression that their distribution author actually built it.
In fact, I *do* build everything on tomsrtbt directly from source, and I
*don't* crib stuff without giving fair credit to the authors and others
who do the work.
I want to prevent people from taking the binary objects and copying them
into their own mini distributions without mentioning where they got them.
I guess one question I have is, suppose I have on my web site:
Where hello.c is a GPL program and a.out is an object file that I
artistically chose to use "-m386 -O2 -fomit-frame-pointer
-fno-strength-reduce", stripped with ELF-Tiny elf object utilities,
used an older GCC to save a few bytes, and cetera.
GPL prevents me from adding restrictions to 'the program'. Well, I am not
restricting it- it is right there- "hello.c" is the program- and I'm glad
to include the *instructions* for how to build it the smallest way- and
the intent of the GPL is clearly to protect the freedom of the source.
What I don't like is other people just copying the *actual binary* without
giving any credit or acknoledgement that *they* don't want to bother to
compile it *themselves*. As long as they mention where they got it, I'm
fine with it.
So, in the case above, am I required not to add the 'give credit and
mention where you got it' clause to the binary object, even if I make the
source code available without any such restriction?
The GPL doesn't really clarify this, it is, after all, aiming at the
source code, and there is probably no consideration that anyone would
*care* about a particular binary compile output.
Now, I can *definitely* apply this restriction to all the scripts and Luas
that I wrote from scratch, *maybe* apply it to the agglomeration and the
arrangement (such as using a very small gzipped root and bzip2ed /usr, is
that copyrightable?) and *maybe* apply it to my versions of the documents
and man pages. It is *possible* that I can apply it to binary objects
where I am making the program *in source* available *without* restriction,
but I dunno from reading the licenses. But I can *ABSOLUTELY* make it
clear that this is my desire, want, and expectation, to the fullest extent
legally available to me. That is what I'm getting at, in my license- "as
far as legally possible, if you reuse ANY of this, you must give credit
to the source".
More information about the License-discuss