Thread: Is anyone working on pg_dump?

One of the major must-fix items remaining on my to-do-for-7.0 list
is making pg_dump smarter about the order to dump stuff in, so as
to avoid problems like table constraints referring to not-yet-
defined functions.

We have talked about this before, and I thought some people had
expressed interest in fixing it, but I haven't seen any results.
Is anyone working on it?

If not, I'm willing to do the quick-and-dirty approach of sorting
the objects by OID.  This could be extended later to allow a
dependency-based sort, but I don't have the time or interest to
attempt that now.

I'd be just as happy to yield the project to someone else though ;-)
        regards, tom lane

Tom Lane writes:

> We have talked about this before, and I thought some people had
> expressed interest in fixing it, but I haven't seen any results.
> Is anyone working on it?

That was me, but not for 7.0. Feel free to try the oid thing, if it works
well, I might not have to do my thing so early.


-- 
Peter Eisentraut                  Sernanders väg 10:115
peter_e@gmx.net                   75262 Uppsala
http://yi.org/peter-e/            Sweden

Tom Lane wrote:
> 
> One of the major must-fix items remaining on my to-do-for-7.0 list
> is making pg_dump smarter about the order to dump stuff in, so as
> to avoid problems like table constraints referring to not-yet-
> defined functions.
> 
> We have talked about this before, and I thought some people had
> expressed interest in fixing it, but I haven't seen any results.
> Is anyone working on it?

I had expressed interest in working on this, but have run out of
spare time. Please feel free.

-- 

Mark Hollomon
mhh@nortelnetworks.com
ESN 451-9008 (302)454-9008

I wrote:
> One of the major must-fix items remaining on my to-do-for-7.0 list
> is making pg_dump smarter about the order to dump stuff in, so as
> to avoid problems like table constraints referring to not-yet-
> defined functions.

I spent a couple evenings working on this, and have made good progress;
but looking at how much I've changed and how much remains, I'm forced
to realize that this is "too big a change for beta".  It requires a
major restructuring of pg_dump.  Since we don't have regression tests
for pg_dump, I think the risk of breaking something is too high for
this phase of the release cycle.

I plan to set the unfinished code aside for now, and come back to it
early in the 7.1 cycle.  We'll have to live with the ordering issue
for another release.
        regards, tom lane