On Mon, Mar 27, 2023 at 06:07:09PM -0700, Peter Geoghegan wrote: > I see a bug on HEAD, following yesterday's commit 0276ae42dd. > > GetWALRecordInfo() will now output the value of the fpi_len variable > before it has actually been set by our call to XXXX. So it'll always > be 0.
Indeed, good catch. It looks like I was not careful enough with the block controlled by XLogRecHasAnyBlockRefs(). -- Michael
signature.asc
Description: PGP signature
