Hi Guix! seg 22 jun 2026 às 23:21:48 (1782181308), [email protected] enviou: > Hi Greg, > > Greg Hogan <[email protected]> skribis: > > > How is "legally significant" being used? > > The link to the GNU maintainers manual is meant to explain what is meant > by “legally-significant”. But you’re right that it is written with > copyright assignment in mind, which can be confusing. > > Perhaps we can drop the link? The “rule of thumb” that follows explains > what it’s about; we can still expound on it if something is missing. > WDYT? >
AFAIUI, "legally significant" is somewhat misleading in that it supposes that the significance is given by law in a clear cut fashion that could be thought as boolean. That is not the case however. A 3-line haiku poem is copyrightable, but a 20-line package simple definition is probably not. The main criteria not being the number of lines, but the freedom of expression in each case. On the haiku side freedom is absolute, on the package definition case it is more like form filling and there's probably no freedom of expression whatsoever (except, perhaps, for package description). So this boolean actually hides a complex sociopolitical question which, IMO, we shouldn't try to legislate a priori on in the absense of a clear legal criteria. Instead, it seems better to leave this ambiguity to be overcome in time by our experiences and those of others. This would be our distributed political duty: on each case to evaluate if some proposal could be harming the interests and rights of others. The questions we should be asking ourselves are: would I be offended if this came from me and attribution or license terms were not followed? Is my perspective on this matter shared by others? Were there any similar cases before? > Disclosure is necessary to determine whether the criteria of item #2 > (“Contribution acceptance”) apply. > > André, Cayetano, and others suggested an ‘Assisted-by’ tag. However I > view it as promotion of the vendor without any technical usefulness, > since the code generation process is not reproducible. > > Instead I would prefer plain text disclosure, a checkbox in the pull > request template, or a tag like ‘LLM-Assistance: debugging’. > I have to take that back. Indeed my sugestion would lead to a popularity contest-like data agregate that would probably be seen as a recomendation pool of sorts, which goes against the spirit of this GCD. Your suggestion is way better and will provide information on what kind of uses are the main candidates for replacement as guix own tools. Cheers!
