[repository migration]

Daniel Kochmański daniel at turtleware.eu
Tue Mar 14 06:08:23 UTC 2023


------- Original Message -------
On Tuesday, March 14th, 2023 at 02:25, Elias Mårtenson <lokedhs at gmail.com> wrote:


> That does sound like a nice thing to add. Is there a standard clause for that? Surely some lawyer somewhere has looked at it?

If we trust Microsoft to be a judge in their own case, then
it is fair use. Some people share this perspective while some
other does not. 
Note that MS does not train copilot on its own closed source code, so if it is taking only small excerpts from the code that have no real value as they claim then this is inconsequential. To me it is simply disrespectful and scummy; not that it is the first time this company tries to work against FOSS.

As far as I'm aware there is a case in court over that. I hope that they win but I personally don't have this kind of time.

> 
> Den tis 14 mars 2023 06:56Anton Vodonosov <avodonosov at yandex.ru> skrev:
> 
> > Are you sure CoPilot violates LGPL?
> > 
> > I mean if author is not happy with this type of use, it may be better to have an explicit clause in the license. For example declaring ML models trained on the licensed code base as derived work, that must be made available to others for free, under the same license.

All kinds of licenses require attribution (including MIT). LGPL requires share-alike. Since their self proclaimed (not tested) argument is that it is a "fair use" then they argue that copyright does not apply here, so licenses and attribution may be simply ignored. Very convenient if you ask me.

Either way even if we had added a license that we do not allow ML training they would gladly ignore it and carry on. That said I'm not against training ML models if they uphold what is expected from the licensee.

Best regards,
Daniel


> > 
> > (in addition to migration of the repository)
> > 
> > 13.03.2023, 22:13, "Daniel Kochmański" <daniel at turtleware.eu>:
> > 
> > > Dear All,
> > > 
> > > I've wanted to do it after the upcoming release but it's months already and it doesn't seem to be much closer than before.
> > > 
> > > Due to GitHub/Microsoft taking our work without proper attribution and not respecting the license (with CoPilot)
> > > we're moving the repository to a non-profit hosting Codeberg. The new home of the repository is:
> > > 
> > > https://codeberg.org/McCLIM/McCLIM
> > > 
> > > The repository on GitHub is archived and available so we won't break existing urls.
> > > 
> > > During the discussion on our IRC channel some questions emerged:
> > > 
> > > - why not gitlab.common-lisp.net?
> > > 
> > > According to some users it is not very reliable - 2fa, repository
> > > limit per user, servers are down from time to time.
> > > 
> > > - what good will that do? people will mirror on github anyway.
> > > 
> > > It is a very sad state of affairs that we can't stand up against
> > > techno-bullies who take from us :| but not doing anything would
> > > be even worse because that would be a silent consent.
> > > 
> > > (with-jokes (:ignorable t)
> > > 
> > > - it learns like human so it is not a violation!
> > > 
> > > yada yada yada :)
> > > 
> > > - but but what about the stars?
> > > 
> > > I'm afraid that our inflated ego may not survive this.
> > > 
> > > )
> > > 
> > > With this humorous accent I wish you all the best and I hope to see you during upcoming ELS in Amsterdam. :)
> > > 
> > > Best regards,
> > > Daniel
> > > 
> > > 
> > > --
> > > Daniel Kochmański ;; aka jackdaniel | Przemyśl, Poland
> > > TurtleWare - Daniel Kochmański | www.turtleware.eu
> > > 
> > > "Be the change that you wish to see in the world." - Mahatma Gandhi



More information about the mcclim-devel mailing list