hcurti@exa.unicen.edu.ar writes:
> PostgreSQL version: 8.4.17
> After statments triggers on child tables are sometimes executed and
> sometimes are not. As I far as I could see, when the after trigger is
> defined in only one of the tables it may not get executed, wether when it is
> defined in every inherited table it does, but
> which one is undetermined.
This test case appears to work the way you're expecting in 9.0 and
later. I didn't immediately find a mention of such a change in the
commit logs; perhaps it got fixed as a side-effect of the changes that
moved trigger processing into ModifyTable plan nodes. Anyway, I doubt
we'd consider changing trigger behavior in 8.4.x at this late date.
You should update to a newer release series if this is a problem for you.
regards, tom lane