Re: [VOTE] Require monotonically increasing snapshot timestamps in v4

[Aihua Xu]

+1 (non-binding) Thanks Aihua On Oct 7, 2026, at 5:58 PM, Junwang Zhao <zhjw...@gmail.com> wrote:  +1 (non-binding) Regards Junwang Zhao On Thu, Oct 8, 2026 at 05:25 Ryan Blue <b...@apache.org> wrote: +1 (binding) On Wed, Oct 7, 2026 at 1:57 PM Yufei Gu <flyrain...@gmail.com> wrote: +1(binding) Yufei On Wed, Oct 7, 2026 at 1:55 PM Péter Váry <peter.vary.apa...@gmail.com> wrote: > > +1 > > Matt Topol <zotthewiz...@gmail.com> ezt írta (időpont: 2026. okt. 7., Sze, 22:30): >> >> +1 (binding) >> >> On Wed, Oct 7, 2026, 4:03 PM Neelesh Salian <neeleshssal...@gmail.com> wrote: >>> >>> +1 (non-binding) >>> >>> On Thu, Oct 8, 2026 at 03:23 Steven Wu <stevenz...@gmail.com> wrote: >>>> >>>> Hi, >>>> >>>> I propose requiring each new snapshot in a v4 table to have a timestamp-ms strictly greater than its parent snapshot’s timestamp on the same branch. Root snapshots are unconstrained, and v1–v3 behavior remains unchanged. >>>> >>>> This ensures snapshot timestamps align with snapshot order. This change would resolve time travel issues caused by out-of-order snapshot timestamps. This can also enable a row timestamp metadata column inherited from the snapshot timestamp (a separate follow-up v4 spec change). >>>> >>>> Spec PR: #16294 >>>> Reference implementation: #16293 >>>> >>>> Please vote in the next 72 hours: >>>> >>>> [ ] +1 Adopt the requirement >>>>  [ ] +0 No preference >>>>  [ ] -1 Do not adopt the requirement (please explain) >>>> >>>> Thanks, Steven >>>> >>>> >>>> Discussion thread: https://lists.apache.org/thread/cln7xtjy9kw2096gzqlg5c4gg8b9td10