Re: [PATCH v4 2/3] docs: define policy limiting the inclusion of generated files

["Daniel P. Berrangé" <berrange@redhat.com>]

On Fri, Jun 06, 2025 at 08:26:37AM +0200, Markus Armbruster wrote:
> Daniel P. Berrangé <berrange@redhat.com> writes:
> 
> > On Thu, Jun 05, 2025 at 12:38:09PM +0100, Peter Maydell wrote:
> >> On Thu, 5 Jun 2025 at 11:52, Markus Armbruster <armbru@redhat.com> wrote:
> >> > +At times contributors may use or create scripts/tools to generate an initial
> >> > +boilerplate code template which is then filled in to produce the final patch.
> >> > +The output of such a tool would still be considered the "preferred format",
> >> > +since it is intended to be a foundation for further human authored changes.
> >> > +Such tools are acceptable to use, provided they follow a deterministic process
> >> > +and there is clearly defined copyright and licensing for their output.
> >> 
> >> For the case where there's a one-off generation step and then the
> >> intent is purely human-authored changes from there onwards, why
> >> do we care whether the tool followed a deterministic process or
> >> not? As long as the copyright/licensing situation is clear and
> >> the submitter has checked tha the generation is what they want,
> >> what does determinism get us?
> >
> > The copyright/licensing is important, but it was trying to say
> > more than that to limit the scenarios in which generated code
> > would be contributed. I think determinism in the tool's operation
> > is valuable, but probably not the key point to get across here.
> >
> > We don't want a free for all in hand editting and then contributnig
> > any auto-generated content. We only want generated content included
> > where it was explicitly intended that it serve as a "template" for
> > human refinement.
> >
> > Determinisism in the sense that if a 2nd person used the same
> > tool to auto-generate the base template for editting, they would
> > be starting from the same place as the original contributior.
> >
> >> As a trivial example, this rules out a hacky one-off python
> >> script that produces output by iterating through a hashtable
> >> if you forgot to add a "sort" to that ordering to make it
> >> deterministic.
> >
> > NB it is trying to say that the way the tool operates is determinstic,
> > not that its output is neccessarily stable wrt things like sorting.
> > ie you can rationalize about what the tool is going to emit, but.
> 
> I think the paragraph's purpose is to clarify "preferred format" is what
> developers want to work with going forward even when the initial
> contribution started with generated contents.
> 
> I think the "deterministic process" clause distracts from this.
> Moreover, it feels largely redundant with the next patch's "Use of AI
> content generators".  Scratch the clause?
> 
>   Such tools are acceptable to use, provided and there is clearly defined
>   copyright and licensing for their output.

Fine for me.

With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|