Re: pg_migrator and handling dropped columns

"Joshua D. Drake" <jd@commandprompt.com> writes:
> On Thu, 2009-02-12 at 13:39 -0500, Tom Lane wrote:
>> a long form only.  And probably not even list it in the user
>> documentation.

> Why wouldn't we want to list it?

Because it's for internal use only.  Although the effect we're
discussing here is relatively harmless, it seems possible that
further down the road we might find a need for hacks that would
render the output entirely unfit for ordinary dump purposes.
I don't see a need to encourage people to play with fire.

It's hardly unprecedented for us to have undocumented internal
options --- there are some in postgres.c for example.
        regards, tom lane