+1 on the PR. Thanks Yong for the fix. Some of these configurations are not temporary, like allowing a custom namespace location.
Yufei On Fri, Feb 6, 2026 at 1:03 PM Michael Collado <[email protected]> wrote: > I've always preferred to have a single set of configuration flags, but the > intention of the behavior-change vs feature config was to allow potentially > risky fixes (see relevant XKCD https://m.xkcd.com/1172/) to be > controllable > by short-lived configuration flags. As long as everyone is ok with the idea > of short-lived configuration flags being treated the same as long-term > feature flags (e.g., turning on/off external catalog credential vending), > then I'm fine with unifying them. > > Mike > > On Fri, Feb 6, 2026 at 9:33 AM Dmitri Bourlatchkov <[email protected]> > wrote: > > > I propose that we start cleaning up this group of flags, promoting all > > the valid ones to standard, permanent configuration flags. > > > > > > +1 to that proposal > > > > Cheers, > > Dmitri. > > > > On Fri, Feb 6, 2026 at 9:23 AM Alexandre Dutra <[email protected]> > wrote: > > > > > Hi Yong, > > > > > > Thanks for initiating this discussion! > > > > > > I have some concerns regarding this specific category of configuration > > > flags. Introduced in [1124], their original purpose appears to have > > > been preventing unexpected behavioral changes in Polaris – even when > > > the older behavior was incorrect. > > > > > > However, I believe we now have clearer evolution guidelines and better > > > methods for communicating breaking changes to our users. It's also > > > important, I think, to never expose a flag that users could leverage > > > to enable buggy behavior. Finally, these flags were explicitly > > > designated as temporary workarounds, yet they are still present, > > > nearly a year later. > > > > > > I propose that we start cleaning up this group of flags, promoting all > > > the valid ones to standard, permanent configuration flags. > > > > > > I don't think they should be documented as they exist today. > > > > > > Thanks, > > > Alex > > > > > > [1124]: https://github.com/apache/polaris/pull/1124 > > > > > > On Fri, Feb 6, 2026 at 4:03 AM Yong Zheng <[email protected]> wrote: > > > > > > > > Hello, > > > > > > > > We are missing doc for > > > > > > https://github.com/apache/polaris/blob/main/polaris-core/src/main/java/org/apache/polaris/core/config/BehaviorChangeConfiguration.java > > . > > > Some of the parameters are really handy and being implicitly set by > > > OpenCatalog as well. Finding the behavior different between Polaris and > > > OpenCatalog and digging through code to find the missing configurations > > is > > > not very fun. I am wondering if we should add these valid > configurations > > to > > > our public page. Once concern from community is "those flags have been > > > there for a long time now and were neither removed nor promoted to > stable > > > flags". > > > > > > > > What do your guys think? They are still valid configuration and there > > is > > > no sign/plan to remove afaik. Should we add them to avoid the code > > digging? > > > > > > > > Thanks, > > > > Yong Zheng > > > > > >
