On 08/25/2011 06:05 PM, Jim Nasby wrote:
> On Aug 24, 2011, at 7:43 PM, Josh Berkus wrote:
>> On 8/23/11 1:30 PM, Andrew Dunstan wrote:
>>> Attached is an undocumented patch that allows pg_restore to omit
>>> post-data items or omit all but post-data items. This has been discussed
>>> before, and Simon sent in a patch back on 2008, which has bitrotted
>>> some. I'm not sure why it was dropped at the time, but I think it's time
>>> to do this. This patch relies on some infrastructure that was added
>>> since Simon's patch, so it works a bit differently (and more simply).
>> If it's not clear from Andrew's description, the purpose of this patch
>> is to allow dividing your pgdump into 3 portions:
>>
>> 1. schema
>> 2. data
>> 3. constraints/indexes
>>
>> This allows users to implement a number of custom solutions for ad-hoc
>> parallel dump, conditional loading, data munging and sampled databases.
>> While doing so was possible before using the manifest from pg_restore
>> -l, the manifest approach has been complex to automate and relies on
>> obscure knowledge.
>>
>> I have immediate production use for this patch and may be backporting it.
> FWIW, I got around this by writing a perl script that calls pg_dump -s and watches for the end of table create
statements(IIRC it specifically looks for the first CREATE INDEX). The advantage to that approach is that you don't
haveto first create a custom format dump and then run pg_restore against that.
>
Well, notwithstanding my well known love of perl, that strikes me as
spending a pound to save a penny. And custom format dumps rock ;-) Also,
your recipe above is buggy, BTW. A CREATE INDEX statement might well not
be the first item in the post-data section.
But we could also add these switches to pg_dump too if people feel it's
worthwhile. I haven't looked but the logic should not be terribly hard.
cheers
andrew