Re: Large block size problems and notes...

Sean Chittenden <sean@chittenden.org> writes:
> 2) While testing the above, I noted that 65K blocks fail and 32K is as
> big as PostgreSQL can handle, for one reason or another.

15-bit offsets in page item ids.

> but, it seems as though the test at the bottom of
> RelationGetBufferForTuple():
> should be changed to:
>         if (len > MaxTupleSize)

No, it shouldn't.  That test would simply be redundant.
        regards, tom lane