Python documentation in DocBook

[Anton Vredegoor]

On 14 Nov 2002 18:45:31 +0100, martin at v.loewis.de (Martin v. Loewis)
wrote:

>In addition to Michael's comments (you need to be able to talk to the
>submitter), there is also the issue of copyright. The PSF currently
>discusses requiring formal papers for contributors, which would raise
>the barrier even more. These procedures are needed to legally maintain
>ownership of Python. Otherwise, some contributor (anonymous or not in
>his contribution) could claim that we have stolen his software, and
>that make an end to Python as free software. Or some organization
>could claim that the contributor has stolen their software and
>contributed it to Python.

Of course the PSF has to be careful to not accept potential poisonous
contributions but it is the responsibility of the PSF to thoroughly
check legal problems and the contributor should not be burdened with
these matters. If the contributor contributes code using a BSD license
for the code, the code should be compatible to the Python license or
else the Python license itself would not be compatible to BSD.

>This is, unfortunately, a requirement that springs up when you develop
>valuable free software: there are always people who will take
>advantage of you, unless you are careful.

Yes you should be careful, but be careful by checking the code and
it's license. If you start assuming people to be taking advantage
unless they can proove that they are not you miss a lot of chances. 

This in no way precludes you from deciding to not use the code because
you don't trust it. Its just that you should first accept the code and
then check if its any good instead of checking the author. The author
could be using a pseudonym for example, that's not the PSF's concern. 

The PSF's concern is only to check if the code is valid and not
stolen. If there are concerns about the code being possibly stolen,
the code contribution process could consist of publishing the code in
a newsgroup comp.lang.python.incoming or something like that, and
people could inspect it for possible legal problems. 

I don't think algorithms are copyrightable so the only possible check
would be that the contributions are written in Python and show no
signs of unpythonic coding which could hint at the code being possibly
translated from an other language. Remains the problem of some former
employee of a company releasing the company's *Python* code without
permission from the company, which would make a clean room
implementation of the code necessary if the python had already
swallowed it. I trust the Python community to be cautious enough to
sense these kinds of problems and profit from the added transparency
more than the possible damage would cost. 

This situation is analogous to the "safety and reliability" issue
discussed in another thread, where some people would rather code in
languages having always the same type of object bound to a certain
name once this name is initialized with that type of object. There is
much to say in favor of giving up on direct security by early fixing
of types, and receiving a gain in total security because of the added
transparency.

>So anonymous contributions are clearly out of question.

I hope I gave an example of someone not being able to contribute as
you asked about in a previous post, although it's a situation
occurring only for the very few people that care about these matters
:-)

>I don't quite understand your concern: Is it that you need to identify
>yourself to SourceForge? In that case, there is not much we can do,
>except for sharing our believe that SourceForge will operate sensitive
>with your private data, and not reveal it to anybody.

128 bits of type checking is not pythonic and will severely limit the
set of possible interactions with your clients. I am not discussing
data: I am discussing protocols.

Regards,

Anton.


"Si non e vero e ben trovato" (old Dutch saying)