[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Encoding the name in the file contents



Jeff Licquia writes:
 > On Wed, 2002-07-24 at 14:56, Walter Landry wrote:
 > > So let me get this straight.  Pristine LaTeX would have, within it, a
 > > mechanism for checking whether a particular file is "blessed" by the
 > > LaTeX project.  Ideally, it could check digital signatures.  md5sums
 > > might be a simpler way to do it.  Anyone is allowed to remove or alter
 > > this or any other feature, the only restriction being that you can't
 > > call the "kernel" LaTeX anymore.
 > 
 > Mostly.  I wouldn't limit this ability to LaTeX only; "fooTeX" could
 > also check for "standard fooTeX" if it wanted.  Also, the particular
 > implementation cited was for modules to declare for themselves what
 > standard they belonged to, instead of relying on a canonical list.  This
 > might be important; although Debian can assume the presence of md5sum or
 > gpg, I doubt the LaTeX Project can.

from my point of view this is a correct summary.  But i would like to note
that the file is not "blessed" by the LaTeX Project (i think feeling this
feeling is the root of some of the resentments). We don't decided which files
are part of ULL (ie "blessed") this decided simply by using LPPL so ULL grows
or changes by the way the LaTeX community contributes further work under this
license.

So it is NOT me or David or anybody else from The LaTeX Team that controls an
this: the terms of LPPL control it as any work under LPPL will be on a LaTeX
system (but not on a fork on) load the sameset of macros if

\usepackage{foo} 

is requested in a document (and "foo" is part of ULL, ie generally available
and under LPPL)



Technically the LaTeX project can't rely on something like md5sum because
there is no sensible way to record and collect the checksum information as we
talk about many many independent works that form the ULL.  The is because ULL
defines itself via the LPPL usage. Equally important perhaps more so is the
ability to do this check at runtime in an simple and fast manner and as we
can't change the underlying binary (at least not if it is TeX).  We therefore
have to build the check inside the LaTeX format itself. But that would be
possible.

If it would be done via something like \NeedsTeXFormat{latex2e} then LPPL
would need to state

that in case of modification and distribution

 - you either rename your work with respect to its loading name for LaTeX (ie
   if you want to keep "\NeedsTeXFormat{latex2e}"). That way it would become a
   further set of definitions adding to the LaTeX Language (LL) and if that
   derived work is put under LPPL also to ULL

 - or that you change to \NeedsTeXFormat{sniffenlatex} if your work is
   intended for a "nonLaTeX" fork in which case you could keep the name.

 > > If this is what the proposal is, then it sounds perfectly DFSG free to
 > > me.  It sounds like a bit of work for the LaTeX people to get the
 > > verification stuff working, but nothing insurmountable.  md5sum is in
 > > the public domain, so there are no licensing problems there.  It
 > > sounds like a technical solution to a legal problem.

This wouldn't work, but in principle something could be made to
work it can however not be made to really do checks on checksums and the like
so it has to rely on the license requesting it.

Henning and Boris just nicely sumarised the argument.


 > I think we can all agree that Debian and LaTeX have some fundamental
 > differences of opinion on things.  We're hoping that the most important
 > of these (namely, license and DFSG coherence) can be resolved; I
 > wouldn't bet any money on the rest.

I don't that that "The Debian People" or "The LaTeX Mafia" exists as such. But
anyway, lets try to get the (...) done, then I bet you some money on the rest
Jeff :-)

cheers
frank


-- 
To UNSUBSCRIBE, email to debian-legal-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



Reply to: