On Thu, Apr 24, 2025, at 2:13 AM, Peter Green wrote: >> >> If preferred, a variant of the proposed changes with a default of "no" would >> also be possible > > I think the default (for bin packages) should be not to generate a multi-arch: > field at all. > > This is behaviorally equivalent to multi-arch: no, but it IMO has different > implications, it implies "noone has thought about multi-arch for this package" > rather than "sometime has thought about multi-arch for this package and > rejected it"
Thanks for the quick feedback! Helmut Grohne raised a similar objection on IRC, and like I noted in unblock request, that works just as well. Having "no" as default is easiest, having no default (which like you say, is "behaviorally equivalent") means a bit more code changes, but can be supported as well.
