On 2/16/19 5:57 AM, Andres Freund wrote: > On 2019-01-05 13:19:20 -0500, Stephen Frost wrote: >> Greetings, >> >> * Peter Eisentraut (peter.eisentraut@2ndquadrant.com) wrote: >>> On 12/12/2018 05:31, Robert Haas wrote: >>>> Most of the features I've been involved in removing have been >>>> deprecated for 5+ years. The first release where this one was >>>> deprecated was only 2 years ago. So it feels dramatically faster to >>>> me than what I think we have typically done. >>> >>> I was just looking this up as well, and I find it too fast. The >>> nonexclusive backups were introduced in 9.6. So I'd say that we could >>> remove the exclusive ones when 9.5 goes EOL. (That would mean this >>> patch could be submitted for PostgreSQL 13, since 9.5 will go out of >>> support around the time PG13 would be released.) >> >> I don't agree with either the notion that we have to wait 5 years in >> this case or that we've only had a good alternative to the exclusive >> backup mode since 9.5 as we've had pg_basebackup since 9.1. > > I don't agree with a general 5 year deprecation window either. But it > seems pretty clear that there's no majority for removing exclusive > backups in v12. I think it'd be good to make the warning about its > impending death more explicit, but otherwise mark this CF entry either > as rejected or returned with feedback.
I think there is support for the patch in PG13 so I was planning to move it out of the March CF to the first PG13 CF as soon as the app will allow it, i.e., once there is only a single open CF.
Agreed, and I think we should also update the documentation for 12 with the suggested more explicit mention of the deprecation.