There is currently no one pushing for import. So I close the task.

Import/Export will (probably be implemented implemented in next CERN branch.

Basic plan:
---------------

Format:
xml files will be structured in a way they could contain as many items as necessary, mentionning to which group they belong.
They won't contain item id. They wont contain user_id and group_id but user unix names and group unix names. Id are site-specific, they should stay that way.
The structure will have to include for each item every necessary info (CC and whatever)

Import method:
Import will be done by adding xml files to load on a specific place in the server.
This should be done occasionally, run by the site admin -- but it could be automatized easily.
A backend script will parse (using perl XML modules) the file and will handle by himself the insert. No mail notification will be sent, no action will be done on the items, considering we are simply importing data of already performed actions. Not acting on items.
Ideally, this should also support bugzilla xml export.

Export method:
Like for the digest, users, via the web interface, will be able to select a list of items (list of items or list of criteria that will be used to define the items). Once satisfied with their list, they will submit the list, this will be stored in the database, and a cronjob will make the backend building the xml file in a specific directory, available via apache.
The idea of letting the frontend doing would pose too many problems: it wont scale if someone wants to work with more than 50 items, it will make the server collapse if persons does plenty of time the same request at the same time, it will make complicated to deal with server load (not making easy to let that job being done by another machine).
Later, it will be possible to imagine allow people to ask for a specific xml export to be done everyweek at a specific hour, or whatever.