Re: Non-LLM example where we do not in practice use original training data

On Tue, 06 May 2025 at 17:12:12 +0200, Simon Josefsson wrote:
>Simon McVittie <[email protected]> writes:
>> Debian is unusual in the way we interpret our mission statement as
>> extending to everything we distribute being Free, not just our
>> executable code.
>
>I don't think Debian is perfectly consistent in applying that principle:
>for example, the text of the Developer Certificate of Origin (DCO) is
>included in Debian packages (in 'main') and has a clearly non-free
>license, and IIRC sometimes not even in debian/copyright.

I think this is seen as part of making a wider exception for licenses 
and similar legal texts. The text of the GPL has similar wording, and we 
certainly can't be a legally-valid Linux distribution without shipping a 
copy of that.

>There are
>other examples historically of similar content too, e.g., IETF RFCs and
>Unicode tables.

I believe project policy is currently that IETF RFCs under the "old" 
non-Free license, even if they are only in source packages, are a DFSG 
violation (although I'm sure there are plenty of undiscovered bugs for 
this topic, and I have no particular interest in proactively looking for 
them).

My understanding is that Unicode tables are under a Free license, and at 
least some packages go to significant effort to ensure that we have a 
preferred form for modification for their content (for example src:glib2.0 
has a copy of the subset of unicode-data necessary to regenerate its 
internal tables, because the version of Unicode that it implements is 
part of its external API and should not be arbitrarily changed 
downstream, but we cannot guarantee that it will always be perfectly in 
sync with the current version of src:unicode-data in Debian).

     smcv