Hi all,

I have posted some examples, along with some notes about them at <http://spdx.org/wiki/openlogic-spdx-10-beta-examples>. The examples are intended to conform to the 1.0 beta version of the spec except that we used sha-256 checksums -- rather than sha-1 -- to identify the files.

I was not able to figure out how to add that page to the examples sandbox. (Perhaps i do not permission to do that? ) Would someone with more knowledge of (or permissions with) the wiki do that for me?

Comments and feedback are welcome.

Peter Williams
<http://openlogic.com>

This is good. It can start some discussion on the standard.

First, one question:

I scanned the file for zlib and I found some issues with it, but I
think are worth discussing:

1. Some files do not contain a license, yet they are marked as one:

dmg@i:/tmp/zlib-1.2.5$ more contrib/minizip/zip.c
/* zip.c -- IO on .zip files using zlib
Version 1.1, February 14h, 2010
part of the MiniZip project - (
http://www.winimage.com/zLibDll/minizip.html )

Copyright (C) 1998-2010 Gilles Vollant (minizip) (
http://www.winimage.com/zLibDll/minizip.html )

Modifications for Zip64 support
Copyright (C) 2009-2010 Mathias Svensson ( http://result42.com )

For more info read MiniZip_info.txt

Changes
Oct-2009 - Mathias Svensson - Remove old C style function prototypes
Oct-2009 - Mathias Svensson - Added Zip64 Support when creating new
file archives
Oct-2009 - Mathias Svensson - Did some code cleanup and refactoring
to get better overview of some functions.
Oct-2009 - Mathias Svensson - Added zipRemoveExtraInfoBlock to
strip extra field data from its ZIP64 data
It is used when recreting zip archive
with RAW when deleting items from a zip.
ZIP64 data is automaticly added to
items that needs it, and existing ZIP64 data need to be removed.
Oct-2009 - Mathias Svensson - Added support for BZIP2 as
compression mode (bzip2 lib is required)
Jan-2010 - back to unzip and minizip 1.0 name scheme, with
compatibility layer

*/


------------
2. Some files refer to zlib.h as the file with a license. Now, if the
SHA1 of the file does not change, I would presume (as a user) that I
don't need to scan it again, which is good. But what if zlib.h
changes? Would it be useful in the SPDX to
use a "reference" field to denote such a thing?

---------
3. Is it the same to include a license than to refer to a license?

---
4. In some files the zlib iicense varies slightly:


This software is provided 'as-is', without any express or implied
warranty. In no event will the author be held liable for any damages
arising from the use of this software.

and in others

This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.

--dmg


On Wed, Sep 29, 2010 at 12:52 PM, Philip Odence
<podence@...> wrote:

I moved it to
Home » Wiki » Software Package Data Exchange (SPDX) » Spec
Development » Sandbox For Sharing Examples, Ideas, Etc.
Not sure if it way my knowledge or permissions or both, but anyway, it's
there.
Good stuff, Peter.



On Sep 29, 2010, at 3:45 PM, Peter Williams wrote:

Hi all,

I have posted some examples, along with some notes about them at
<http://spdx.org/wiki/openlogic-spdx-10-beta-examples>.  The examples
are intended to conform to the 1.0 beta version of the spec except that
we used sha-256 checksums -- rather than sha-1 -- to identify the files.

I was not able to figure out how to add that page to the examples
sandbox.  (Perhaps i do not permission to do that? )   Would someone
with more knowledge of (or permissions with) the wiki do that for me?

Comments and feedback are welcome.

Peter Williams
<http://openlogic.com>
_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx


_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx

--
--dmg

---
Daniel M. German
http://turingmachine.org

On 9/29/10 2:32 PM, dmg wrote:

This is good. It can start some discussion on the standard.

First, one question:

I scanned the file for zlib and I found some issues with it, but I
think are worth discussing:

1. Some files do not contain a license, yet they are marked as one:

We assume any that file that does not contain explicit license info and does not match any of the open source in our database is licensed under the declared license of the project. In this case the Zlib license.

2. Some files refer to zlib.h as the file with a license. Now, if the
SHA1 of the file does not change, I would presume (as a user) that I
don't need to scan it again, which is good. But what if zlib.h
changes? Would it be useful in the SPDX to
use a "reference" field to denote such a thing?

I think this is outside the scope of the spdx proper. Many of the decisions about what licenses govern a file will be made on criteria other than an explicit license declaration, direct or indirect. For example, some part of a file might be matched against a database of open source and that open source file might have a license associated with it.

In the short term this could be handled as comment on the file object. It might be an interesting follow on project to create an extension to allow expressing the decision criteria for why a particular license was chosen.

3. Is it the same to include a license than to refer to a license?

We treat those the same. This is a policy issue to be worked out between the producer and the consumers of the spdx file. I think the spec should avoid specify the copyright/license analysis process. Spdx should just provide a way to express the results of such an analysis.

4. In some files the zlib iicense varies slightly:


This software is provided 'as-is', without any express or implied
warranty. In no event will the author be held liable for any damages
arising from the use of this software.

and in others

This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.

This also feels like a policy issue to me. We treat those as the same.

Peter Williams
<http://openlogic.com>

Thanks Peter for your clarifications.

I think this shows, that the ones creating the files will be _making_
decisions. In this case, several have been made:

1. Files without a license share the license of the project
2. If a file A specifies that its license is in B, then license(A) == license(B)
3. Even thought there is no perfect textual comparison of the license
(aside from whitespace) the licenses have been considered to be
equivalent.

These are very good reasons why standardizing text of licenses by
inclusion seems to me like a bad idea.

---dmg

On Thu, Sep 30, 2010 at 9:06 AM, Peter Williams
<peter.williams@...> wrote:

On 9/29/10 2:32 PM, dmg wrote:

This is good. It can start some discussion on the standard.

First, one question:

I scanned the file for zlib and I found some issues with it, but I
think are worth discussing:

1.  Some files do not contain a license, yet they are marked as one:

We assume any that file that does not contain explicit license info and does
not match any of the open source in our database is licensed under the
declared license of the project.  In this case the Zlib license.

2. Some files refer to zlib.h as the file with a license. Now, if the
SHA1 of the file does not change, I would presume (as a user) that I
don't need to scan it again, which is good. But what if zlib.h
changes? Would it be useful in the SPDX to
use a "reference" field to denote such a thing?

I think this is outside the scope of the spdx proper.  Many of the decisions
about what licenses govern a file will be made on criteria other than an
explicit license declaration, direct or indirect.  For example, some part of
a file might be matched against a database of open source and that open
source file might have a license associated with it.

In the short term this could be handled as comment on the file object. It
might be an interesting follow on project to create an extension to allow
expressing the decision criteria for why a particular license was chosen.

3. Is it the same to include a license than to refer to a license?

We treat those the same.  This is a policy issue to be worked out between
the producer and the consumers of the spdx file.  I think the spec should
avoid specify the copyright/license analysis process.  Spdx should just
provide a way to express the results of such an analysis.

4. In some files the zlib iicense varies slightly:


  This software is provided 'as-is', without any express or implied
  warranty.  In no event will the author be held liable for any damages
  arising from the use of this software.

and in others

  This software is provided 'as-is', without any express or implied
  warranty.  In no event will the authors be held liable for any damages
  arising from the use of this software.

This also feels like a policy issue to me.  We treat those as the same.

Peter Williams
<http://openlogic.com>
_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx

--
--dmg

---
Daniel M. German
http://turingmachine.org

On 9/30/10 11:57 AM, dmg wrote:

Thanks Peter for your clarifications.

I think this shows, that the ones creating the files will be _making_
decisions.

I completely agree. I think anyone that has actual tried to analyze a package for copyright/license info knows that a lot of judgment calls are required.

In this case, several have been made:

1. Files without a license share the license of the project
2. If a file A specifies that its license is in B, then license(A) == license(B)

I would say that as license(A) = license-specified-by(B). For example, the text of GPL v3, <http://www.gnu.org/licenses/gpl.html>, is licensed under terms quite different from GPL. So if license(A) -> B where B is a file containing just the text of the GPL then license(A) = GPL but license(B) = "Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed."

3. Even thought there is no perfect textual comparison of the license
(aside from whitespace) the licenses have been considered to be
equivalent.

This is the only sane thing to do. Unfortunately, there are situations in which reasonable people could disagree about whether two license texts are really the same license or not.

These are very good reasons why standardizing text of licenses by
inclusion seems to me like a bad idea.

Here i disagree. I think standardizing some license texts is a Good Thing. No one will be force to reference those standard licenses. If you find a license that you believe is materially different from the all the texts in the public repo that license can be included in the spdx file as a non-standard license. Having a set of licenses with standardized names allows much more efficient communication and greater interoperability.

The standard should be updated to allow the license text to be included in all situations. Even for standard licenses. That way an spdx producer could include the variations found, even if the producer considers them materially the same.

Peter

---dmg

On Thu, Sep 30, 2010 at 9:06 AM, Peter Williams
<peter.williams@...> wrote:

On 9/29/10 2:32 PM, dmg wrote:

This is good. It can start some discussion on the standard.

First, one question:

I scanned the file for zlib and I found some issues with it, but I
think are worth discussing:

1. Some files do not contain a license, yet they are marked as one:

We assume any that file that does not contain explicit license info and does
not match any of the open source in our database is licensed under the
declared license of the project. In this case the Zlib license.

2. Some files refer to zlib.h as the file with a license. Now, if the
SHA1 of the file does not change, I would presume (as a user) that I
don't need to scan it again, which is good. But what if zlib.h
changes? Would it be useful in the SPDX to
use a "reference" field to denote such a thing?

I think this is outside the scope of the spdx proper. Many of the decisions
about what licenses govern a file will be made on criteria other than an
explicit license declaration, direct or indirect. For example, some part of
a file might be matched against a database of open source and that open
source file might have a license associated with it.

In the short term this could be handled as comment on the file object. It
might be an interesting follow on project to create an extension to allow
expressing the decision criteria for why a particular license was chosen.

3. Is it the same to include a license than to refer to a license?

We treat those the same. This is a policy issue to be worked out between
the producer and the consumers of the spdx file. I think the spec should
avoid specify the copyright/license analysis process. Spdx should just
provide a way to express the results of such an analysis.

4. In some files the zlib iicense varies slightly:


This software is provided 'as-is', without any express or implied
warranty. In no event will the author be held liable for any damages
arising from the use of this software.

and in others

This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.

This also feels like a policy issue to me. We treat those as the same.

Peter Williams
<http://openlogic.com>
_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx

In my opinion, the problem with allowing "user judgement" in included
license variability can lead to disagreements of what a license really
is, or even worse, misunderstanding of what the license of a file is.

Say hypothetically, you read a license and for you it is zlib, and for
me it is not, and I prefer to refer to it as a zlib-variant, because
for me the differences are strong enough to worry.

I would prefer that there was a single place at the beginning of the
SPDX file where such two variants of the license are located, and then
I can look at it and decide if it is equal or not. Rather than
trusting your judgement.

Perhaps I am just beating a dead horse, and nobody really cares about
such differences (think MIT/X11 and BSD-variants not this zlib
example).

--dmg

On Thu, Sep 30, 2010 at 1:19 PM, Peter Williams
<peter.williams@...> wrote:

3. Even thought there is no perfect textual comparison of the license
(aside from whitespace) the licenses have been considered to be
equivalent.

This is the only sane thing to do.  Unfortunately, there are situations in
which reasonable people could disagree about whether two license texts are
really the same license or not.

These are very good reasons why standardizing text of licenses by
inclusion seems to me like a bad idea.

Here i disagree.  I think standardizing some license texts is a Good Thing.
 No one will be force to reference those standard licenses.  If you find a
license that you believe is materially different from the all the texts in
the public repo that license can be included in the spdx file as a
non-standard license.  Having a set of licenses with standardized names
allows much more efficient communication and greater interoperability.

--
--dmg

---
Daniel M. German
http://turingmachine.org

Two more things about the zlib example:

1. The license of the ada subdirectory is GPLv2+ not, GPLv2.

2. There is another interesting example, which is labelled BSD-3 in
the SPDX. Same issues regarding this than the variability of the zlib
license apply here.

/*
* match.S -- optimized version of longest_match()
* based on the similar work by Gilles Vollant, and Brian Raiter, written 1998
*
* This is free software; you can redistribute it and/or modify it
* under the terms of the BSD License. Use by owners of Che Guevarra
* parafernalia is prohibited, where possible, and highly discouraged
* elsewhere.
*/

--
--dmg

---
Daniel M. German
http://turingmachine.org

PErhaps the solution is to have a judgement field, that indicates if
the license is matched perfectly, or a decision was made.

I also think it would be very useful to extract the license statement
of file, and save it. As tools get better then can concentrate on the
analysis of such,k particular for the extraction of copyright
information.

--
--dmg

---
Daniel M. German
http://turingmachine.org

On 9/30/10 2:54 PM, dmg wrote:

PErhaps the solution is to have a judgement field, that indicates if
the license is matched perfectly, or a decision was made.

I also think it would be very useful to extract the license statement
of file, and save it. As tools get better then can concentrate on the
analysis of such,k particular for the extraction of copyright
information.

A judgment is always made. Even if the file says "licensed under the terms of the BSD License", you have to decide if you believe that or if you believe they copied the file from a GPL licensed project and stripped the original license header.

Peter
<http://openlogic.com>

On Thu, Sep 30, 2010 at 1:45 PM, dmg<dmg@...> wrote:

Two more things about the zlib example:

1. The license of the ada subdirectory is GPLv2+ not, GPLv2.

2. There is another interesting example, which is labelled BSD-3 in
the SPDX. Same issues regarding this than the variability of the zlib
license apply here.

/*
* match.S -- optimized version of longest_match()
* based on the similar work by Gilles Vollant, and Brian Raiter, written 1998
*
* This is free software; you can redistribute it and/or modify it
* under the terms of the BSD License. Use by owners of Che Guevarra
* parafernalia is prohibited, where possible, and highly discouraged
* elsewhere.
*/

On Wed, Sep 29, 2010 at 1:32 PM, dmg<dmg@...> wrote:

This is good. It can start some discussion on the standard.

First, one question:

I scanned the file for zlib and I found some issues with it, but I
think are worth discussing:

1. Some files do not contain a license, yet they are marked as one:

dmg@i:/tmp/zlib-1.2.5$ more contrib/minizip/zip.c
/* zip.c -- IO on .zip files using zlib
Version 1.1, February 14h, 2010
part of the MiniZip project - (
http://www.winimage.com/zLibDll/minizip.html )

Copyright (C) 1998-2010 Gilles Vollant (minizip) (
http://www.winimage.com/zLibDll/minizip.html )

Modifications for Zip64 support
Copyright (C) 2009-2010 Mathias Svensson ( http://result42.com )

For more info read MiniZip_info.txt

Changes
Oct-2009 - Mathias Svensson - Remove old C style function prototypes
Oct-2009 - Mathias Svensson - Added Zip64 Support when creating new
file archives
Oct-2009 - Mathias Svensson - Did some code cleanup and refactoring
to get better overview of some functions.
Oct-2009 - Mathias Svensson - Added zipRemoveExtraInfoBlock to
strip extra field data from its ZIP64 data
It is used when recreting zip archive
with RAW when deleting items from a zip.
ZIP64 data is automaticly added to
items that needs it, and existing ZIP64 data need to be removed.
Oct-2009 - Mathias Svensson - Added support for BZIP2 as
compression mode (bzip2 lib is required)
Jan-2010 - back to unzip and minizip 1.0 name scheme, with
compatibility layer

*/


------------
2. Some files refer to zlib.h as the file with a license. Now, if the
SHA1 of the file does not change, I would presume (as a user) that I
don't need to scan it again, which is good. But what if zlib.h
changes? Would it be useful in the SPDX to
use a "reference" field to denote such a thing?

---------
3. Is it the same to include a license than to refer to a license?

---
4. In some files the zlib iicense varies slightly:


This software is provided 'as-is', without any express or implied
warranty. In no event will the author be held liable for any damages
arising from the use of this software.

and in others

This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any damages
arising from the use of this software.

--dmg


On Wed, Sep 29, 2010 at 12:52 PM, Philip Odence
<podence@...> wrote:

I moved it to
Home » Wiki » Software Package Data Exchange (SPDX) » Spec
Development » Sandbox For Sharing Examples, Ideas, Etc.
Not sure if it way my knowledge or permissions or both, but anyway, it's
there.
Good stuff, Peter.



On Sep 29, 2010, at 3:45 PM, Peter Williams wrote:

Hi all,

I have posted some examples, along with some notes about them at
<http://spdx.org/wiki/openlogic-spdx-10-beta-examples>. The examples
are intended to conform to the 1.0 beta version of the spec except that
we used sha-256 checksums -- rather than sha-1 -- to identify the files.

I was not able to figure out how to add that page to the examples
sandbox. (Perhaps i do not permission to do that? ) Would someone
with more knowledge of (or permissions with) the wiki do that for me?

Comments and feedback are welcome.

Peter Williams
<http://openlogic.com>
_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx


_______________________________________________
Spdx mailing list
Spdx@...
https://fossbazaar.org/mailman/listinfo/spdx

--
--dmg

---
Daniel M. German
http://turingmachine.org

--
--dmg

---
Daniel M. German
http://turingmachine.org