I originally had an objection to sending email after only 1 day of a package being stuck.
Consider packages that take longer than a day to build - these will be invalid (out of date), so you'll mail before they are ready AFAICS. I don't think proposed-migration can actually tell if a package is out of date because it is still building or because it's failed to build, either, so as it stands I don't think this is fixable. That's why we currently wait at least 3 days before mailing.
On reflection, though, I think I'm okay with merging this change and we'll see if that is a problem in reality.
Would you please comment the code where you calculate last_due? That is quite difficult to reason about and it would benefit from some explanation up front. You can just push that directly yourself.
I originally had an objection to sending email after only 1 day of a package being stuck.
Consider packages that take longer than a day to build - these will be invalid (out of date), so you'll mail before they are ready AFAICS. I don't think proposed-migration can actually tell if a package is out of date because it is still building or because it's failed to build, either, so as it stands I don't think this is fixable. That's why we currently wait at least 3 days before mailing.
On reflection, though, I think I'm okay with merging this change and we'll see if that is a problem in reality.
Would you please comment the code where you calculate last_due? That is quite difficult to reason about and it would benefit from some explanation up front. You can just push that directly yourself.