Am Donnerstag, den 19.12.2019, 00:13 +0000 schrieb Simon Riggs: > So the consensus is for a more-specifically named facility. > > I was aiming for something that would allow general SELECTs to run > with a > snapshot that can see uncommitted xacts, so making it a SRF wouldn't > really > allow that.
There's pg_dirtyread() [1] around for some while, implementing a SRF for debugging usage on in normal circumstances disappeared data. Its nice to not have worrying about anything when you faced with such kind of problems, but for such use cases i think a SRF serves quite well.
As an example of an SRF for debugging purposes, sure, but then we already had the example of pageinspect, which I wrote BTW, so wasn't unfamiliar with the thought.
Note that pg_dirtyread has got nothing to do with the use cases I described.