Merged. https://github.com/apache/polaris/pull/2369
On 2025/08/19 18:59:21 Dmitri Bourlatchkov wrote: > I updated PR [2369] to use PURGE_VIEW_METADATA_ON_DROP as the feature flag > name. > > Please review, approve or comment. As I noted previously, I'm open to other > naming suggestions, but this name appears to be the middle ground based on > prior discussion. > > [2369] https://github.com/apache/polaris/pull/2369 > > Thanks, > Dmitri. > > On Mon, Aug 18, 2025 at 4:54 PM Dmitri Bourlatchkov <di...@apache.org> > wrote: > > > PURGE_VIEW_METADATA_ON_DROP (suggested by Yun) looks like a good middle > > ground to me. > > > > It still has "PURGE", which relates to the underlying machinery in the > > Polaris server > > that will be triggered to delete related files in storage, and at the same > > time it has > > "METADATA", which hopefully clarifies the intended behaviour for users. > > > > WDYT? > > > > Thanks, > > Dmitri. > > > > > > On Mon, Aug 18, 2025 at 4:27 PM Russell Spitzer <russell.spit...@gmail.com> > > wrote: > > > >> I like DELETE_METATDATA on drop, but i'm not opposed to purge, if you > >> don't > >> have strong feelings either way i'm fine with it staying Purge. > >> > >> On Mon, Aug 18, 2025 at 3:14 PM yun zou <yunzou.colost...@gmail.com> > >> wrote: > >> > >> > +1 on updating the name to be more accurate — I was initially confused > >> > when I first saw the field name. > >> > > >> > The new proposed name 'DELETE_VIEW_METADATA_ON_DROP' sounds much more > >> > clear to me, or if > >> > we would like to stay more consistent on the term used, maybe we can > >> > call it 'PURGE_VIEW_METADATA_ON_DROP'? > >> > > >> > Best Regards, > >> > Yun > >> > > >> > > >> > On Mon, Aug 18, 2025 at 7:34 AM Russell Spitzer > >> > <russell.spit...@gmail.com> wrote: > >> > > > >> > > My only thought on this is that usually I consider a purge a drop of > >> the > >> > > "data" but in this case there isn't actually any data, it's only > >> > metadata. > >> > > > >> > > I know we are kind of copying the old hive/Iceberg naming here where > >> the > >> > > catalog "metadata" was considered independent from the on > >> > > disk metadata. So it feels right in that history, I'm just wondering > >> if > >> > we > >> > > can start using a better vocab to describe what's happening. So I have > >> > > no issues with going ahead with this, I hope we can find some better > >> > > vocabulary around this since the idea of dropping a view with "purge" > >> > > still sounds a bit odd to me. > >> > > > >> > > So overall a +1, but I think we should consider not even having a > >> "purge" > >> > > verb for views with a possibly different behavior than drop. > >> > > > >> > > On Mon, Aug 18, 2025 at 7:24 AM Robert Stupp <sn...@snazy.de> wrote: > >> > > > >> > > > +1 > >> > > > > >> > > > On Sat, Aug 16, 2025 at 6:22 AM Dmitri Bourlatchkov < > >> di...@apache.org> > >> > > > wrote: > >> > > > > > >> > > > > Hi All, > >> > > > > > >> > > > > I propose to add [1] a new feature flag PURGE_VIEWS_ON_DROP to > >> allow > >> > > > > dropping views when DROP_WITH_PURGE_ENABLED is false (default). > >> > > > > > >> > > > > The default value of PURGE_VIEWS_ON_DROP is true to match prior > >> > > > behaviour. > >> > > > > > >> > > > > Any concerns? > >> > > > > > >> > > > > [1] https://github.com/apache/polaris/pull/2369 > >> > > > > > >> > > > > Thanks, > >> > > > > Dmitri. > >> > > > > >> > > >> > > >
