+1 (binding) The proposed V3 behavior would already be a lot more flexible than what most engines support in the industry today. It is also not covered by the SQL standard, so there is no need to overcomplicate the spec without actual use cases.
- Anton ср, 23 квіт. 2025 р. о 10:27 Ryan Blue <rdb...@gmail.com> пише: > +1 (binding) > > On Wed, Apr 23, 2025 at 8:39 AM Fokko Driesprong <fo...@apache.org> wrote: > >> +1 (binding) >> >> Kind regards, >> Fokko >> >> Op wo 23 apr 2025 om 03:08 schreef Gang Wu <ust...@gmail.com>: >> >>> +1 (non-binding) >>> >>> On Wed, Apr 23, 2025 at 4:42 AM Prashant Singh <prashant010...@gmail.com> >>> wrote: >>> >>>> +1 (non-binding) >>>> >>>> Best, >>>> Prashant Singh >>>> >>>> On Tue, Apr 22, 2025 at 2:55 AM Eduard Tudenhöfner < >>>> etudenhoef...@apache.org> wrote: >>>> >>>>> +1 >>>>> >>>>> On Tue, Apr 22, 2025 at 7:31 AM Jean-Baptiste Onofré <j...@nanthrax.net> >>>>> wrote: >>>>> >>>>>> +1 (non binding) >>>>>> >>>>>> Regards >>>>>> JB >>>>>> >>>>>> On Mon, Apr 21, 2025 at 11:20 PM Ryan Blue <rdb...@gmail.com> wrote: >>>>>> > >>>>>> > Hi everyone, >>>>>> > >>>>>> > I’d like to vote on the spec changes in PR 12841. This is a small >>>>>> change that makes handling default values for structs much easier. >>>>>> Initially, we allowed both a struct and its fields to have default >>>>>> values, >>>>>> but the values could conflict. For instance, ADD COLUMN point struct<x >>>>>> int >>>>>> default 0, y int default 0> default struct(-1, -1). >>>>>> > >>>>>> > The fix is to always track default values at the field level and >>>>>> allow only null or null-null for the struct level defaults. That makes >>>>>> the >>>>>> feature more predictable because the struct’s default never needs to be >>>>>> modified or have field-level changes applied (i.e. removing field y or >>>>>> adding field z). >>>>>> > >>>>>> > In addition, I want to mention that this is not a one-way decision. >>>>>> We can always allow the struct-level default to differ later, if we have >>>>>> use cases in which a missing struct needs to have a different default >>>>>> than >>>>>> missing fields. >>>>>> > >>>>>> > Please vote in the next 72 hours: >>>>>> > >>>>>> > [ ] +1 Add these changes to the spec >>>>>> > [ ] +0 >>>>>> > [ ] -1 I have questions and/or concerns >>>>>> > >>>>>> > Thanks, >>>>>> > >>>>>> > Ryan >>>>>> >>>>>
