(1) pgplot licence vs other; (2) NR (fwd)

Boud Roukema boud w astro.uni.torun.pl
Śro, 31 Mar 2004, 13:09:46 CEST


witam,

Michał wrote:

> Boud Roukema wrote:
>
> >>what is the purpose of GPL? creating better, community driven software.
> >>imho it is more than a license, it is a philosophy. your "software" just
> >
> >
> > It is not philosophy. It is politics, because it relates to the power that
> > some people have over other people, and the power that people do *not*
> > have over other people.
> >
> > The intention is to minimise the power that people have over others to
> > the minimum necessary - in the context of computer code.
> >
>
> It depends what you mean by "politics" and "philosophy" I believe. in
> fact it is both. but developers seem to prefer the term "philosophy"
> since it is more positive in meaning.

i guess this may be an anglophone/francophone difference. Personally i
prefer precision rather than political correctness.

> >>  do not fit GPL. it is a set of mathematical operations.
> >>how the hell do you want people to use it/trace bugs/improve when you
> >>only provide hardly readable code??????
> >>
> >>we could argue about the quality of the code but I do not think will do
> >>any good.
> >
> >
> > The point is not to argue about the quality of the code.
> >
> > The whole point of the GNU GPL is that concrete discussion of the
> > code can happen according to rules which encourage improvement of the code.
> >
> > Without distribution under GNU GPL, it is difficult to get the *feedback*
> > that someone has the opinion that the code is "hardly readable".
> >
>
> I agree.

:)

> but without providing high-quality code and docs it is hard to
> get useful feedback.

It's sufficient to improve by the same fraction.

Consider a code of low initial quality p(0) = 0.3 with low docs
quality q(0)=0.3, versus code with high initial quality p'(0)= 1.0 and
high initial docs quality q'(0) = 1.0.  In the first case there is on
average one iteration of feedback each week which yields moderate
improvement in the combined quality  x = pq of 10% (alpha = 1.1), in
the second case the initial version is only distributed after one year,
and then gets feedback yielding improvements of 10%.

p(0)=q(0)=0.3
p'(0)=q'(0)=1.0

x(t) = p(0)q(0) alpha^n   (n= weeks)
x'(t) = p'(0)q'(0) alpha^(max(0,n-52))

which gives for t \ge 52:

x(t)= 12.8 alpha^(n-52)
x'(t)= 1.0 alpha^(n-52)

Conclusion: many iterations through a positive feedback loop starting
from an imperfect beginning can give a better result than waiting a long
time for a "perfect" beginning to be ready.


> > On the other hand, if there is a delay of between several days and
> > a few weeks for having the right to use the code, and then a further
> > delay of unknown length to redistribute modified versions, the cooperation
> > between X, Z, W and U is unlikely to happen, especially given that Y
> > feels unhappy with the code and may influence the delays.
> >
> I can bet that Kriss would be more that happy to accept any help with
> porting his software to other languages. Have you tried to chat with him
> about licence issues?

If you wished to be a neutral, third party moderator to discuss with a
certain person regarding certain software, for a publicly archived
discussion (like this one) regarding porting of cosmologically useful
software to a version distributed under the GNU GPL or a compatible
licence, and if you convinced that person and other people interested
that you would be a good moderator (meaning that you clarify the
discussion, stop different people from flaming or threatening each other,
ask them to apologise or find other ways of getting people to cool
off if they start flaming, etc.), i would be happily accept this.

My prediction is that this would be not be accepted by some of the people,
but i would be glad to be wrong.


> As I still am quite sure about there is no licence that suits scientific
> software and gpl is not the best. that is why some people refuse
> gpl-ing. I myself would be very happy having a nicer licence for sci soft.

Personally i'm not convinced of the need. I'm aware that if you spend
a lot of time on software and don't write many publications, you need to
get credit for the software. The GNU GPL already guarantees that the
original author's name remains, and IMHO the cosmo community is small
enough that people writing, or more importantly, maintaining good software
do get credit. Just because it's not formally part of these strange
formulas attempting to measure scientific productivity in PL, does not
mean the credit is not there.

However, you want a modified licence, so please have a look at:

http://www.gnu.org/licenses/gpl-faq.html#ModifyGPL

What do think of a licence which is identical to the GNU GPL, but
we remove the preamble (unless FSF agrees that we may include it in our
modified version after discussion), we modify the end-wording,
and we add a term/condition like:

SSL-13.0  - proposal for Scientific Software Licence-term/condition 13.0
----------------------------------------------------------------------
 13. Any publication in a recognised scientific journal which makes
use of this software must contain the acknowledgment: "Use has been
made of PROGRAM-NAME which can be obtained at PROGRAM-URI", where
PROGRAM-NAME is the name of the Program and PROGRAM-URI is a Universal
Resource Identity showing where the full machine-readable source code
of the Program can be downloaded from."
----------------------------------------------------------------------


and then write to  licensing at gnu.org  to ask for advice.

My guess is they would only complain about the compatibility question:

http://www.gnu.org/licenses/gpl-faq.html#WhatIsCompatible


In which case i would suggest in addition:

SSL-2.1  - proposal for Scientific Software Licence-term/condition 2.1
----------------------------------------------------------------------
identical with GNU GPL v2 term/condition 2. except for the following:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
"But when ... wrote it. "
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

becomes

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
"But when you distribute the same sections as part of a whole which is
a work based on the Program, the distribution of the whole must be on
the terms of either this License, whose permissions for other
licensees extend to the entire whole, and thus to each and every part
regardless of who wrote it, or of the GNU General Public License, as
published by the Free Software Foundation (either version 2 of the
License, or (at your option) any later version) whose permissions for
other licensees extend to the entire whole, and thus to each and every
part regardless of who wrote it.

If you apply the GNU General Public License rather than this License to
the whole, then you are requested (but not required) to include a file
called README.acknowledgments in the machine-readable source code
which includes the text of term/condition 13. below, with the phrase

"Any publication ... acknowledgment:"

replaced by

"It is kindly requested (but not legally required) that any
publication in a recognised scientific journal which makes use of this
software should contain the acknowledgment:"
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

----------------------------------------------------------------------


In short, this means that if an SSL protected package is combined with
a non-GNU GPL package, then:

- if the other package is also under SSL, the new one can also be SSL
- if the other package is GNU GPL, then the new one *must* be GNU GPL
(condition 2 of GNU GPL), but the user is *requested* to add a file
converting term/condition 13 into a README.acknowledgments file which
*requests* the user to write an acknowledgment
- if the other package is under some other licence, it may or may not
be possible to distribute the combined package, it depends on that other
licence.


IMHO, this is as strong as we could get while retaining GNU GPL
compatibility. If we try to make the request into a requirement, then
my worry is it would be GPL incompatible.

In fact, compatibility is one of the big issues of the GNU GPL. It
enables modules to be combined together in ways far beyond what the
original authors ever could do alone. Remember: there are 6 billion
of us. That's a lot.

Anway, you wanted a licence: what do you think of this one?

If we go through a few iterations, with a bit of feedback :), then i'd
be happy to write to licensing at gnu.org, point them to our
discussion, and see what they think.


A simpler alternative to SSL-13.0 + SSL-2.1 would be:

SSL-13.1  - proposal for Scientific Software Licence-term/condition 13.1
----------------------------------------------------------------------
 13. It is kindly requested (but not legally required) that any
publication in a recognised scientific journal which makes use of this
software should contain the acknowledgment:  "Use has been
made of PROGRAM-NAME which can be obtained at PROGRAM-URI", where
PROGRAM-NAME is the name of the Program and PROGRAM-URI is a Universal
Resource Identity showing where the full machine-readable source code
of the Program can be downloaded from."

If you distribute this Program as part of a whole which is a work
based on the Program, and you apply the GNU General Public License
(GNU GPL) to that whole, then you are kindly requested to copy this
term/condition 13. to a file called README.acknowledgments (if it is
not already present) and include it in the machine-readable source
code. Since further distribution of the whole would occur under the
GNU GPL, there would be no further obligation to include the file, only
a kind request.
----------------------------------------------------------------------


pozdr
boud


 



Więcej informacji o liście Cosmo-torun