"Usama Munir" <usama.munir@enterprisedb.com> writes:
> I guess the way forward for me would be to crawl back in my corner,
> write up a mini - spec of how i intend to implement it and get back to
> you guys.
Well, the *first* thing to do is read pg_dump for awhile. Until you've
grokked what it does to support multiple backend versions and how it
handles inter-object dependencies (dump ordering problems), you won't be
able to write a spec that has any credibility. In particular, the
business about breaking circular dependency loops is something I have
no idea how to handle in a simple "dump library" API.
regards, tom lane