Open isedwards opened 5 years ago
In our last project meeting we briefly discussed open source licences.
@githubetna suggested AGPL @Steve-Palmer suggested LGPL @isedwards mentioned that we may have multiple licences...
For example:
Also, some time ago we received advice to adopt a permissive licence until the final formal decision of licence is made. Starting with a permissive licence makes it easier to change licence in the future (as indicated by the arrows on the diagram in the comment above).
This is a hard one to give a definitive opinion on. By nature I lean towards collaboration via a permissive license. But then again there is still a small but finite risk that some actors with an exploitative bent may attempt to appropriate our code for their own purposes. Do the existing licences, or OpenWIS, protect us against patent trolls?
Provided that we have that level of protection, I would feel more confident about backing open licensing.
Will
On Tue, 25 Aug 2020 at 22:00, Ian Edwards notifications@github.com wrote:
In our last project meeting we briefly discussed open source licences.
@githubetna https://github.com/githubetna suggested AGPL https://opensource.org/licenses/AGPL-3.0 @Steve-Palmer https://github.com/Steve-Palmer suggested LGPL https://opensource.org/licenses/lgpl-license @isedwards https://github.com/isedwards mentioned that we may have multiple licences...
For example:
- The API could be released under a permissive licence to encourage maximum possible adoption
- Processes coming from existing software may continue to be made available under their exisiting licence (e.g. GPL3 for R-Instat)
Also, some time ago we received advice to adopt the permissive BSD licence until the final formal decision of licence is made. Starting with a permissive licence makes it easier to change licence in the future (as indicated by the arrows on the diagram in the comment above).
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/opencdms/opencdms-project/issues/14#issuecomment-679980852, or unsubscribe https://github.com/notifications/unsubscribe-auth/AP55ZGUT6TAXEXKV7VEUCV3SCORUHANCNFSM4HY5HP3Q .
-- Dr. William J Wright Climate Data-Management Consultant Co-chair on Data - Commission for Climatology, WMO e: wjwright1385@gmail.com m: +61 (0) 457 915 375 444 Healesville-Kooweerup Rd, Healesville Victoria 3777 Australia [image: Please consider the environment before printing]
I think it is a very small risk that anyone would sell a commercial service using OpenCDMS code either without attribution or passing it off as their own – the market for such a product is small and already well-populated with commercial offerings; there is no significant commercial benefit in denying attribution, and some benefit in claiming use of OpenCDMS. It does change the business model slightly from selling a product to selling support for use (similar to the business model for Redhat Linux). You can see a similar situation with the recent Microsoft publicity that the Edge browser has been completely changed from proprietary code to use the Google Chromium code.
Of more concern is the reputational risk that people will modify the OpenCDMS code without feeding back or checking, and get something wrong.
If ever something got so bad that we need to start issuing legal warnings, then the OpenWIS Association is set up to deal with this. It is incorporated under Belgian law, which explicitly does not recognise software patents (and thus means no one can patent OpenWIS Association property under any other patent regime because it would clearly not be original). The Association has access to IPR lawyers in the organisations which own the Association (including Michael Robbins in the Met Office, who has guided me on a lot of this work).
I am mainly concerned that we positively want people to use the OpenCDMS code and processes as “best-practice” so for the time being, I am keen to start with permissive licenses, especially to encourage use by the commercial providers, so their products become hybrid rather than the current closed systems.
We put Climsoft under GPL3 as a strong statement, which was right for the transition we needed to move the Climsoft project forward at a time when the decision was either to abandon or go open-source, but GPL3 is probably too strong for where we are now with OpenCDMS. Hence the idea of using MIT, BSD or LGPL which do not have the strong inheritance restriction, and multiple licenses may help on the different parts.
OK, thanks Steve. I had an idea that OpenWIS did indeed have IP protections, but thanks for confirming. In that case I am happy to agree on a permissive license for the software. .
As for Creative Commons - I spent a good proportion of two years of my life trying to convince the new breed of Bureau Senior Executives that we should go that way for Bureau products and services - which was consistent with Australian Government policy to make data and information collected at public expense as broadly accessible as possible to encourage innovation and entrepreneurship - and consistent with long-standing Bureau practice. It is also in spirit consistent with WMO Policy. But they could not, or would not, accept those arguments. Funnily enough, most of those Senior Execs have since been moved on!
Cheers
Will
On Wed, 26 Aug 2020 at 19:00, Steve Palmer notifications@github.com wrote:
I think it is a very small risk that anyone would sell a commercial service using OpenCDMS code either without attribution or passing it off as their own – the market for such a product is small and already well-populated with commercial offerings; there is no significant commercial benefit in denying attribution, and some benefit in claiming use of OpenCDMS. It does change the business model slightly from selling a product to selling support for use (similar to the business model for Redhat Linux). You can see a similar situation with the recent Microsoft publicity that the Edge browser has been completely changed from proprietary code to use the Google Chromium code.
Of more concern is the reputational risk that people will modify the OpenCDMS code without feeding back or checking, and get something wrong.
If ever something got so bad that we need to start issuing legal warnings, then the OpenWIS Association is set up to deal with this. It is incorporated under Belgian law, which explicitly does not recognise software patents (and thus means no one can patent OpenWIS Association property under any other patent regime because it would clearly not be original). The Association has access to IPR lawyers in the organisations which own the Association (including Michael Robbins in the Met Office, who has guided me on a lot of this work).
I am mainly concerned that we positively want people to use the OpenCDMS code and processes as “best-practice” so for the time being, I am keen to start with permissive licenses, especially to encourage use by the commercial providers, so their products become hybrid rather than the current closed systems.
We put Climsoft under GPL3 as a strong statement, which was right for the transition we needed to move the Climsoft project forward at a time when the decision was either to abandon or go open-source, but GPL3 is probably too strong for where we are now with OpenCDMS. Hence the idea of using AGPL or LGPL which do not have the strong inheritance restriction, and multiple licenses may help on the different parts.
Regards Steve
From: WilliamJWright notifications@github.com Sent: 25 August 2020 17:10 To: opencdms/opencdms-project opencdms-project@noreply.github.com Cc: Palmer, Steve steve.palmer@metoffice.gov.uk; Mention < mention@noreply.github.com> Subject: Re: [opencdms/opencdms-project] Open Source License and Contributor's License Agreement (#14)
This is a hard one to give a definitive opinion on. By nature I lean towards collaboration via a permissive license. But then again there is still a small but finite risk that some actors with an exploitative bent may attempt to appropriate our code for their own purposes. Do the existing licences, or OpenWIS, protect us against patent trolls?
Provided that we have that level of protection, I would feel more confident about backing open licensing.
Will
On Tue, 25 Aug 2020 at 22:00, Ian Edwards <notifications@github.com mailto:notifications@github.com> wrote:
In our last project meeting we briefly discussed open source licences.
@githubetna https://github.com/githubetna suggested AGPL https://opensource.org/licenses/AGPL-3.0 @Steve-Palmer https://github.com/Steve-Palmer suggested LGPL https://opensource.org/licenses/lgpl-license @isedwards https://github.com/isedwards mentioned that we may have multiple licences...
For example:
- The API could be released under a permissive licence to encourage maximum possible adoption
- Processes coming from existing software may continue to be made available under their exisiting licence (e.g. GPL3 for R-Instat)
Also, some time ago we received advice to adopt the permissive BSD licence until the final formal decision of licence is made. Starting with a permissive licence makes it easier to change licence in the future (as indicated by the arrows on the diagram in the comment above).
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/opencdms/opencdms-project/issues/14#issuecomment-679980852>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AP55ZGUT6TAXEXKV7VEUCV3SCORUHANCNFSM4HY5HP3Q>
.
-- Dr. William J Wright Climate Data-Management Consultant Co-chair on Data - Commission for Climatology, WMO e: wjwright1385@gmail.commailto:wjwright1385@gmail.com m: +61 (0) 457 915 375 444 Healesville-Kooweerup Rd, Healesville Victoria 3777 Australia [image: Please consider the environment before printing]
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub< https://github.com/opencdms/opencdms-project/issues/14#issuecomment-680121948>, or unsubscribe< https://github.com/notifications/unsubscribe-auth/AC32BTQABG3VSB36SI7ITYDSCPO6XANCNFSM4HY5HP3Q>.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/opencdms/opencdms-project/issues/14#issuecomment-680753887, or unsubscribe https://github.com/notifications/unsubscribe-auth/AP55ZGWV2EMH7POWPAAAVR3SCTFMFANCNFSM4HY5HP3Q .
-- Dr. William J Wright Climate Data-Management Consultant Co-chair on Data - Commission for Climatology, WMO e: wjwright1385@gmail.com m: +61 (0) 457 915 375 444 Healesville-Kooweerup Rd, Healesville Victoria 3777 Australia [image: Please consider the environment before printing]
As an open source project, OpenCDMS needs open source license(s) and a contributor's license agreement (CLA). These should both be in place as early as possible.
The OpenWIS Association require that we use licences approved by the Open Source Initiative.
License
Climsoft and R-Instat both make use of the GPL v3 license. Although this is a good choice for desktop applications, it may not be as suitable for OpenCDMS (which is a reference implementation using web technologies that is being constructed as multiple reusable libraries).
I've seen some libraries start life with the GPL v3 licence and then later have to be rewritten from scratch to adopt something more permissive (e.g. MIT, BSD or LGPL).
The Lesser GPL (LGPL) license, for example, still requires anyone who makes changes, and then shares the software with others, to also release their changes under the LGPL licence. But it doesn't force downstream projects to also adopt the same licence.
Also see https://opensource.google.com/docs/thirdparty/licenses/