Stephen Frost <sfr...@snowman.net> writes: > * Tom Lane (t...@sss.pgh.pa.us) wrote: >> We could hard-code a rule like that, or we could introduce a new >> explicit parameter for the maximum cover length. The latter would be >> more flexible, but we need something back-patchable and I'm concerned >> about the compatibility hazards of adding a new parameter in minor >> releases. So on the whole I propose hard-wiring a multiplier of, >> say, 10 for both these cases.
> That sounds alright to me, though I do think we should probably still > toss a CFI (or two) in this path somewhere as we don't know how long > some of these functions might take... Yeah, of course. I'm still leaning to doing that in TS_execute_recurse. regards, tom lane