February 03, 2005
Ben Hinkle wrote:
| I just went looking at the OSI license list and interestingly enough
| I couldn't find one that mentioned "public domain" anywhere. You'd
| think the open source folks would allow the most open license of all.
| oh well. :-(

Usually asking the copyright holder to license their work to you under
the ZLIB license isn't that cumbersome and avoids potential legal quarrels.

http://opensource.org/licenses/zlib-license.php

Thomas


February 03, 2005
Anders F Björklund wrote:
| Thomas Kuehne wrote:
|
|> Details like deadlines etc. will be posted on November 17. in this
|> newsgroup.
|
|
| The website said "February 17", which sounds a little more interesting
|
| ;-)

:-P

Thomas

February 03, 2005
Anders F Björklund wrote:
| Thomas Kuehne wrote:
|
|> Basically there are several issues with the current Phobos:
|>
|> 1) there are core components with unclear/no license details
|
|
| That is a big problem, indeed. Guess I hadn't looked
| at every file... Is there a list compiled somewhere ?
| (license audit of the DMD front-end and Phobos library)

The following are the results of a very brief review and are likely to
be incomplete.

The DMD front-end(dual license GPL / Artisitc) issues:

port.h, mem.c, inifile.c
	copyright header(Digital Mars/Walter Bright) without the
	license details

mem.h
	copyright header(Chromium Communications) without
	license details

gnuc.h, gnuc.c
	neither copyright nor license headers


Phobos is, simply going by the dmd.zip package and ignoring any
newsgroup posts by Walter, non-redistributable (see license.txt).

(As stated earlier, it's difficult do identify the core components and
their dependencies.)

Core components with unclear/missing copyright headers:

internal/arraycast.d
internal/arraycat.d
internal/cast.d
internal/complex.c
internal/dmain2.d
internal/invariant.d
internal/mars.h
internal/memset.d
internal/object.d
internal/switch.d
std/typeinfo/*
std/array.d
std/asserterror.d
std/moduleinit.d
std/outofmemory.d
std/switcherr.d

Core components with problematic licenses (except the general
non-redistributable problem):

internal/qsort.d (Prototronics, licensed to Digitalmars)

In addition several of the core components simply state "public domain".
I'm no legal expert but almost every country has it's own definition of
"public domain". Maybe asking the copyright holders to license under the
~ zlib/libpng terms (http://opensource.org/licenses/zlib-license.php)
will evade potential troubles.

Thomas
February 03, 2005
Thomas Kuehne wrote:

> | I couldn't find one that mentioned "public domain" anywhere. You'd
> | think the open source folks would allow the most open license of all.
> 
> Usually asking the copyright holder to license their work to you under
> the ZLIB license isn't that cumbersome and avoids potential legal quarrels.

Well, if the original code is Public Domain then *anyone* can do that ?
Just add a new copyright statement and the license text, and it's done.

--anders
February 03, 2005
"Anders F Björklund" <afb@algonet.se> wrote in message news:ctt9u4$2rb3$1@digitaldaemon.com...
> Thomas Kuehne wrote:
>
>> | I couldn't find one that mentioned "public domain" anywhere. You'd
>> | think the open source folks would allow the most open license of all.
>>
>> Usually asking the copyright holder to license their work to you under the ZLIB license isn't that cumbersome and avoids potential legal quarrels.
>
> Well, if the original code is Public Domain then *anyone* can do that ? Just add a new copyright statement and the license text, and it's done.
>
> --anders

But the original is forever in the public domain.


February 03, 2005
"Anders F Björklund" <afb@algonet.se> wrote in message news:ctt94n$2q3n$1@digitaldaemon.com...
> Ben Hinkle wrote:
>
>> I just went looking at the OSI license list and interestingly enough I
>> couldn't
>> find one that mentioned "public domain" anywhere. You'd think the open
>> source
>> folks would allow the most open license of all. oh well. :-(
>
> Public Domain has to do with copyright, not *so* much with licensing. (or more specifically that the author gives up any copyright claims)
>
> The problem with public domain is that it is so easy to close again ? Most open source uses copyright to make sure that the rights remain...
>
> But as far as I know, Public Domain can use the "brand" Open Source
> - as long as it is accompanied by the actual source code, of course...
>
> Kind of an interesting meta-question, if the absence of copyright and/or license is a copyright/license in itself? But not for this D newsgroup.
>
> --anders

A license is a set of rules set down by the copyright holder. So if there is no copyright holder the question of license is moot. The software is free to evolve without restriction. Anyone can take it in any direction.

I wonder if anyone has studied open source software from the evolutionary point of view. I remember seeing something like a family tree for Unix during the SCO battle. They were argueing snippets of code get copy/pasted between projects willy nilly and while that's a no-no in copyrighted software (unless the license allows ... umm... offspring) it is perfectly ok to do with public domain code. In some sense one can argue that if we had a robust population of public domain code that the mixing of code bases would result in a much faster development pace and result in a better, more fit, set of code. That's part of the point of open source, I know, but the darn licenses still get in the way IMO. It's like claiming copyright ownership of a fragment of someone's DNA - which is something the medical community is trying to figure out, actually.


February 03, 2005
Thomas Kuehne wrote:

> In addition several of the core components simply state "public domain".
> I'm no legal expert but almost every country has it's own definition of
> "public domain". Maybe asking the copyright holders to license under the
> ~ zlib/libpng terms (http://opensource.org/licenses/zlib-license.php)
> will evade potential troubles.

It the code is in the Public Domain, then there is no copyright holder.

--anders
February 03, 2005
Ben Hinkle wrote:

> I wonder if anyone has studied open source software from the evolutionary point of view. I remember seeing something like a family tree for Unix during the SCO battle. They were argueing snippets of code get copy/pasted between projects willy nilly and while that's a no-no in copyrighted software (unless the license allows ... umm... offspring) it is perfectly ok to do with public domain code. In some sense one can argue that if we had a robust population of public domain code that the mixing of code bases would result in a much faster development pace and result in a better, more fit, set of code. That's part of the point of open source, I know, but the darn licenses still get in the way IMO. It's like claiming copyright ownership of a fragment of someone's DNA - which is something the medical community is trying to figure out, actually. 

All these copyright wars will seem like a light breeze when the real
patent war storm start. Then the implementation won't matter anymore.

To me, it's just another-thing-that-keeps-me-from-actually-coding...
And I just hate that. :-(    http://www.nosoftwarepatents.com/

--anders
February 03, 2005
zwang wrote:
> Interesting!  I was thinking of an entry that qualifies in all three categories -- an obfuscated library that crashes d compilers :)

That is what I usually produce when I'm trying to write serious code, so  I could just take any of my old code I have lying around...

Let's just hope that it works in reverse, so if I try to contribute to the contest I will write the most beautiful code ever :)

/Anders Runesson
February 03, 2005
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Benjamin Herr schrieb am Thu, 03 Feb 2005 13:03:24 +0100:
> Anders F Björklund wrote:
>> Thomas Kuehne wrote:
>>> 4) no support for 64bit platforms
>> 
>> Clearly a "bug" in the current implementation.
>> Or are you referring to actual API changes ?
>
> I tried changing the un-64bit-ish "ulong _d_newarray(...)" to return an Array struct (same size on 32bit platforms even) but it did not work - I think that the DMD generates code assuming that a scalar ulong is returned and not an aggregate type.
>
> However, I do not see how replacing phobos can fix that.

The compiler and the core lib components are rather tightly coupled. Unless on of them starts identifying and tackling the problems no progress will happen.

Thomas


-----BEGIN PGP SIGNATURE-----

iD8DBQFCAjpJ3w+/yD4P9tIRAsoFAJ45m2thbUNEG8w99A69nqtPf1pucgCfWnoG
JkgSlHLBqQKiuFoWL4WdL2Q=
=bIiZ
-----END PGP SIGNATURE-----