I'd be willing to help with the html parsing of the web/db boards too if there is no sql dump available. I still have the "split a bboard thread in messages" code lying around somewhere from the work on the 3.x -> 4 links migration script.
About the poster's identities: wouldn't it be possible to add them as persons (the acs-object with that name), not users? From a short look at the datamodels it seems that nothing should stop us from inserting postings from persons that have no user account. This seems cleaner then generating fake email-adresses IMHO.
And maybe one day there will even be admin pages that allow an administrator to elevate an existing person entry to a user account. This could be of more general interest: imagine a company that stores all it's client contacts as persons in it's OpenACS system, and upon request can create user accounts for their clients by just typing their email address in an admin form. Has anyone done somehting like that already?
Anyway, if a merge is required there is propably not such a big difference between doing it with two user objects or a user and a person object.
