Side effect of person-account split?

[Jeroen Vermeulen]

Hi Stuart,

As the reviewer for your branch I guess I _should_ have seen this 
coming: an import failed because the process lacked rights for the 
Account table.

I think this is because the importer comes across an unknown name/email 
and creates a Person, but the transitional auth-split branch creates an 
Account in the process.  The poimport user gets the right to insert new 
Persons from the "write" group, but it doesn't have the right to insert 
Accounts.

I'll look into why this didn't come up in tests; I remember seeing a 
test but it may be running under the wrong user.  Meanwhile, can we give 
either "poimport" or "write" insert privs on Account?  The latter would 
also fix any other obscure cases of this problem that we might not have 
spotted.  I know we're supposed to be phasing out the "write" group, but 
the implicit creation of accounts should stop with the next phase of the 
auth split anyway.


Jeroen
-------------- next part --------------
An embedded message was scrubbed...
From: root at forster.canonical.com (Cron Daemon)
Subject: Cron <launchpad at forster> nice -16 python2.4
	/srv/launchpad.net/production/launchpad/cronscripts/rosetta-poimport.py
	-q
Date: Wed, 02 Jul 2008 10:45:42 +0100
Size: 3317
Url: https://lists.ubuntu.com/archives/launchpad-users/attachments/20080703/7f2a7942/attachment-0001.eml