I like how matplotlib handles it
https://matplotlib.org/devdocs/devel/contribute.html#generative-ai
This has been inspired by the linux policy itself as well
Also see
https://github.com/PrismLauncher/PrismLauncher/blob/develop/CONTRIBUTING.md
I'd normally be in favour of banning LLM usage entirely, and many
projects have done so with great success, however I fear that banning it
entirely may end up pushing people to use it but hide it, which may be
worse.
LLM output has been becoming more difficult to recognise, it makes less
mistakes, however the mistakes it makes are well hidden and takes effort
to dig out. Having it be declared in the open which code is computer
generated and which code has been carefully written will allow the
maintainer to look sufficiently critically when needed.
I'd be in favour of excluding fully computer generated code. For mixed
code cases or where an LLM has been consulted, this falls into a more
grey area, and it should be declared.
On 25/06/2026 19:43, Even Rouault via Discuss wrote:
Hi,
The most surprising thing is that this thread has not yet been opened
and isn't hundreds of messages long! but this is similar to bug
reporting: "how come nobody has not yet reported this issue? I can't
be the first one to hit it for sure" ;-) I hope people still follow
this list...
Anyway, I let the floor open to anyone who want to talk about they
feel about the subject, what approaches they have taken, their
enthusiasm and/or concerns, if that's something that is just in the
responsibility of each project or if some common positions can emerge,
etc.
Even
_______________________________________________
Discuss mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/discuss
_______________________________________________
Discuss mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/discuss
