On 2015-08-10 11:19:56 -0400, Tom Lane wrote:
> Meh. I think we'd agreed in another nearby thread that pageinspect
> is exactly the sort of thing we don't want in core.
I think we agreed that it shouldn't be included in an initdb'ed
database, but that doesn't preclude pageinspect being an extension in
src/extension or something.
> Even if it were
> in core, I'm dubious that it's a good way to implement the desired
> testing here: the risks of platform-dependent results (endianness,
> varying numbers of tuples per page, etc etc) seem far too high.
That on the other hand I can agree with being a danger.
Greetings,
Andres Freund