Simon Riggs wrote:
> On Thu, 2008-09-11 at 15:25 -0400, Andrew Dunstan wrote:
>
>
>> Great, thanks (and also to Guillaume).
>>
>> It looks to me like the simple way around this issue would be to provide
>> an option to have pg_restore emit:
>> begin; truncate foo; copy foo ... commit;
>>
>> The truncate will be trivial as there won't be any data or indexes at
>> that stage anyway.
>>
>
> Not sure which stage you're talking about. If this is a parallel restore
> and you are running a create in one session and a load in another, then
> ISTM you have no way of knowing that for certain.
>
>
Er, who doesn't know what for certain, exactly? pg_restore will
certainly know that it has created the table in another session and can
thus safely truncate the table in the same transaction as the data load.
cheers
andrew