Re: licenses in generated m4 files

[Guido Draheim]

Peter Simons wrote:
> Alexandre Duret-Lutz writes:
> 
>  > [various points]
> 
> I think it's better to agree to disagree than to continue
> discussing this. IMHO, it is highly unlikely that anyone's
> opinions will be changed no matter how long we continue to
> argue, so I'd like to focus on the solution instead.


So, the discussion has reached a point ;-) - I do not like
to partake in the details since I am opposing the GPL right
in its concept: all the gpl'ed software with an impact in the
university and the industry... has some exception clause
somewhere allowing the free-use that the core gpl prohibits.
However each of this exception is not compatible with the one
from the next gpl-with-exception project making essentially
for a fractionized software world - with the bottom line that
you just can not take over any code from any other gpl'ish
project (in order not to spoil the exception). Somehow I feel
that we face again this gpl idiocy when talking about the
ac-archive license mix. And in a way I understand that RMS
opted for an all-permissive one - as it does not raise more
basic suspicions about the value of the gpl in the first place.

As I am somewhat biased I did choose to stay out of the
discussion in order to allow any technical solution to get
afloat if there would be such which I did doubt from the
beginning based on my understanding of the fractionized-gpl.
On the sfnet branch I did choose to sometimes add an advertiser
to bsd'ish license concepts as it seems that most ac-macro
submissions do have an affiliation with that and only the
requirements of the ac-archive made a gpl restriction on
top of it.

As from technical points we have already used a maintaince
strategy to not update a macro independently but instead
point back to the original author - which does essentially
require a `please keep my name intact` note somehwere - be
it in the macro or elsewhere. That's not the air of the gpl
where a guy may pick up code from multiple places and massage
them to his likings feeling later full responsibility to all
the errors and problems in it. The core autoconf project may
be seen as working like that but it does not extend to the
ac-archive. I feel like the gpl clause for the ac-archive
was only taken as heritage to the core autoconf but it does
not bear real value on hinting people about how things
should be handled. Full stop.


> [...]
> 

As for the current technicality, one might choose to expand
the tooling that it moves a short "copyright" line up to
the header of an assembled aclocal.m4/acinclude.m4 - such
a line would be marked specially in each macro and from
the assembled list (just one lines with a pointer to the
real macro later in the assembly) one can easily see if
they match. That concept moves the decision of good/bad a
mix to the final user of the macros which do have to look
at that anyway when living in a fractionized-gpl world.
Moving/copy it to the top of the file is needed for legal
means as it makes the user more easily `aware` of possible
problems being involved.

Perhaps it would be good to define a special marker as a
technical equivalent of the aclocal exception-from-gpl
that says it is okay to assemble macros in a bigger file
for usage with autoconf. That can be tweaked later into
better tool support of checking whether the resulting
aclocal.m4/acinclude.m4 is good for an autoconf-run or
not. And just ask submitter if we may add that extra
marker - that will lead quicker to a good result than
discussion licensing options and the preference of each
macro writer - just ask "are you fine with an autoconf
exception clause", and the rest can be whatever the
original author feels like - whatever the license is,
all we need is the autoconf-run exception...

cheers,
-- guido                 http://google.de/search?q=guidod
GCS/E/S/P C++/++++$ ULHS L++w- N++@ s+:a d(+-) r+@>+++