On Sat, Jan 18, 2025 at 02:03:33PM +0100, Michail Nikolaev wrote: > I was able to stabilize (I hope so) tests on [0] without any changes to > isolationtester and without "notices N". > > So, I agree, it is better to put it on a shelf now. But a few words in > documentation may be a good idea.
Could you summarize here what you have done to achieve test stabilization in your new patch set posted at [1] without using the proposal of this thread? Perhaps that's OK to leave things as they are, but your proposal here is leading to a more natural definition in terms of how test ordering can be enforced with injection points, leading to a less confusing definition. The analysis of existing tests if this proposal moves forward would be required, of course. [1]: https://www.postgresql.org/message-id/CANtu0ohStYQkJL+_=40odcqu2a4qprm_vmofwkkudhghfgy...@mail.gmail.com -- Michael
signature.asc
Description: PGP signature
