Josh Berkus wrote:
> Tom Lane wrote:
>> Andrew Dunstan <andrew@dunslane.net> writes:
>>> OK, here 'tis.
>>
>> Looks fairly reasonable to me, but of course I haven't tested it.
>
> Well, I have to do a blitz of parallel restores next week, so
> hopefully that will hit any soft spots.
I have a known outstanding bug to do with deadlock from FKs that cross
(i.e. A has an FK that references B, and B has an FK that references A).
The solution to this could be mildly complex, but I have an outline of
it in my head. Workaround: recreate the failed FK at the end of the restore.
The only other reported problem is the one on Unixware to do with
closing the archive. I haven't been able to reproduce it on Linux or
Windows, the two platforms I test on, although it might be fixed by the
patch I just applied.
cheers
andrew