Reported by user in the public discord. It appears they are hitting a bug in the subscription plan cache. Note, query plans don't yet support proper parameters. Hence we must take care to compute the key correctly when caching plans that are inherently parameterized. RLS plans fall into this category, but so do ViewContext views which we are most likely ignoring when computing the query hash.
Reported by user in the public discord. It appears they are hitting a bug in the subscription plan cache. Note, query plans don't yet support proper parameters. Hence we must take care to compute the key correctly when caching plans that are inherently parameterized. RLS plans fall into this category, but so do
ViewContextviews which we are most likely ignoring when computing the query hash.