On Wed, May 27, 2020 at 9:29 AM Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl> wrote:
On Wed, May 27, 2020 at 03:12:44PM +0200, Marius Schwarz wrote:
> Am 27.05.20 um 08:25 schrieb Artur Iwicki:
> > While I understand the mechanism, I think that this needs to be communicated more clearly. I've been a packager for close to 3 years now and I admit until I read this e-mail I wasn't quite sure whether "no updates after EOL" meant "you can't submit stuff to bodhi", or whether it meant "the updates repo is frozen solid, whatever didn't make it in, well shucks". (As we can see, it's the latter.)
>
> I checked that update in question and it was ready 9 days ago, but
> simply noone cared to give it karma.
>
> I think: To make to work effort acknowlaged, maintainers put into theire
> work, all finished/not failed updates should be send to stable when EOL
> is flipped.

Yeah, I think that'd be better then simply dropping them, if it's
reasonably easy to implement. By releasing those updates we make
things a bit nicer for users who are staying on a
now-slightly-but-as-time-goes-on-more-and-more-so out-of-date release,
and we'd be conserving the work of our packagers. But if it would be a
major hassle for infra or releng, then meh.

I think a phased approach would be (hopefully) easy to implement. 

1. Stop builds at <date>
2. Stop submitting updates on <date>+1 week
3. Stop allowing pushes to stable on <date>+2 weeks.

Something like that anyway.

Thanks,
Richard