Re: LaTeX & DFSG
On Thu, 18 Jul 2002, Boris Veytsman wrote:
> Right now I as an end user and developer have the following rights:
> R1. Change the appearance of any document I got by adding the line
> inputting my set of macros to the document.
This was never in contention, and is irrelevant to the freedom of the
> R2. Change the appearance of all documents by (1) using instead of the
> command "latex file" a command "modified-latex file" or (2)
> passing the corresponding options to tex or (3) using my own
> version of tex with different name.
This seems untrue. The current proposal is that in addition to naming the
main invocation command something other than latex, there are a bunch of
files that cannot be changed without renaming. I have no clue how onerous
this is in practice, but it definitely makes modification more difficult
than you state. Perhaps difficult enough that it qualifies as non-free.
> R3. Distribute my macros, modified versions of latex, tex, fonts etc.,
> for profit or not, as long as I take care they will not be
> confused with the original tex, latex and fonts, and (in case of
> latex) impose the same "do not confuse users" restriction on the
> modified product (TeX is trademarked, so this restriction would be
> redundant for it).
This is true for some modifications. The issue under discussion is the
judgement call of whether the "do not confuse users" restriction is
implemented so intrusively that R2 is infringed.
> A1. If I issue the command "latex file", the appearance of the
> resulting document will be exactly the same as intended by the
> author with discrepancy no more than tens of Angstroms.
Sure if "latex" even exists. The modified version latex-local may be all
you have. In that case, you'll need to get the original sources and build
it yourself. Which you could do anyway without the license restriction.
> A2. If I send my document to be latexed to my publisher, colleague,
> friend, the appearance of the document on their desks will be
> exactly same as on mine. This is true whether they use Debian
> Linux, FreeBSD, Windows, Macintosh or Palm Pilot.
Sure, unless they use latex-local for all their processing.
> A3. The propeties A1 and A2 are going to be there whether the document
> is processed today, tomorrow or in any foreseeable future.
With enough caveats that it's not much of an assurance.
> Now what other rights are you talking about? This could not be my
> rights as a user a developer -- I have all that I need.
"All the rights you need" is a scary concept, but I'll let it go. The
rights you list are sufficient to go into Debian, IFF recipients of the
software really have them as you state.
>From the discussion so far, R2 is much more restricted than you state.
Since the assurances you list aren't really assured, why not drop the
restrictions on the rights?
> must be some other rights. Probably *your* rights. I take from the
> discussion that some people here want the right to modify parts of the
> TeX suite while still calling the thing TeX and LaTeX.
Nope. Nobody objects to calling the project or invocation file something
other than latex. We object to the necessity of managing implementation
details like filenames inside our derived work.
If the license clearly states that we're allowed to have modified versions
of any file in the package as long as we don't call the package latex, the
discussion is done. It's free software.
> In other words
> you want the right to change files on your server in such a way that
> "apt-get upgrade" will make my machine produce slightly different
> documents than my colleagues' machines. Worse than that, you want the
> right to dump these changes on other Debian machines -- while I can
> stop using your verson of LaTeX, I cannot presume my colleagues'
> admins are clueful enough to make this decision.
According to R2, that right already exists. latex-deb could change
radically. Latex is guaranteed to be pristine, though it's not guaranteed
to work (or exist) at all.
> Note that it is not a right to try to make a better LaTeX than LaTeX:
Why not? R2 as you state it allows this, as long as it's named ltx++ or
something that isn't latex.
> Some people say they need this right to make (La)TeX more secure.
They're off on a tangent. The right we need is for our users to be able
to change software we distribute however they want.
> the dangers that LPPL tries to prevent are quite real: the story with
> "improved" CM fonts comes to my mind. Again, the people who improved
> CM fonts might think that they made a service to the community; they
> might consider the changes to be small.
Small or large, the freedom to change the behavior is the key.
> Nevertheless they are NOT small for me and I do not want this service.
Nobody forces you to accept this service. Get the original package, or
use a distribution that you trust not to make such a change. The fact
that you don't want this service is fine. The fact that you don't want
people you've never met to offer the service is the very definition of
> You break A1--A3 without meaningfully adding anything to R1--R3.
A1-A3 should be encouraged outside the license, as R2 completely breaks
their enforcement anyway. Freedom is certainly a danger, and it's real.
IMO it's worth the risk.
> You now say that you need a freedom to make 950 g weights and call them
> kilograms. You say that these weights might be better or more secure.
> You even assure me that you might never need to excersize this right or
> that your weights might be 999 g. However, you consider the right to
> distribute altered kilograms very important.
I require this right in order to consider software free. I have this
exact right, in fact, in GNU Units. I don't even have to rename anything
units would still be free if I had to rename it in order to get the new
kilogram definition. units would not be free if I couldn't make it return
950g for a kilogram.
Party C (my downstream recipient), has the same right, and may prefer
1000g units for kilograms. She can change it back and distribute the
result, or she can choose to get the original package. She could even
change it to 1056g. She'll probably also choose a different distribution
from now on unless there really was a good reason for the change.
But it's her choice, not yours and not mine. You can offer your
definition, I can offer mine, but the end choice is made by the user.
> Thanks, but no thanks. I do not want you to have this freedom.
Fair enough. That puts you clearly in the non-free camp. I'm sorry that
you don't trust people to know better than you how they want their
software to behave. You're probably even correct in assuming that some
people are too stupid to have such a right, but removing it is abhorrent
If you're looking for benign examples, imagine a constrained system (think
embedded) that has a homegrown incomplete tinyTeX. They want to use latex
on it, and it just doesn't work out of the box. Are they allowed to
create and ship a tinylatex with modifications that make it kind of work,
but produce vastly different output than regular latex on TeX?
Or maybe someone wants to create a package that simply increases fontsize
of all documents by 50%. They don't want to (and maybe can't) edit the
It seems that both of these are technically allowed, but it may be that
the implementation filename restrictions make it so impractical that it's
not doable without tons of extra work. THAT is the confusion - it's not
disallowed, but it's made way too hard, so it comes down to human
judgement whether it's "free enough".
Mark Rafn email@example.com <http://www.dagon.net/>
To UNSUBSCRIBE, email to firstname.lastname@example.org
with a subject of "unsubscribe". Trouble? Contact email@example.com