Edit: hmmm ... I'm running into some permissions issues. I'm checking with Scott to see if he changed something.

There's a new dump posted.

There are a couple of changes (mostly discussed in the few posts above this one). The biggest one is that post histories are now included! See the readme for details about what all the fields mean.

LastEditorDisplayName isn't in the non-public table unless the user has been deleted (and so LastEditorUserId doesn't exist). Looking the the SO public dumps (at least the Stack Apps dump, since that's the smallest one to deal with), it looks like LastEditorDisplayName is always given, but is always empty!

I see you've removed the reference to LastEditorDisplayName in the README.txt. Can't we instead include all of LastActivityDate, LastActivityUserId and LastActivityDisplayName, in the public dumps? Surely this is public data too!

1. Include the GUID for each revision, in posthistory.xml. This is public information, contained in the URL for "view source" on the list of revisions.
2. Include the Gravatar hash: take the user's email address or last login IP address, as a string, and compute the MD5 hash.

There are two tables in the database. One contains the posts in their current state in html (for fast serving). The other contains an entry for every edit, retag, or other action that can be taken on a post; notably, it contains the markdown source. It shouldn't be too bad to include another file, posthistories.xml, in the public dump. It would make the dump perhaps 50% larger.

A question about the database dumps -- they don't include the edit history of posts. I assume, @Anton, that you have these in the full dump? Was there some reason to censor these?

When you vote to close (or reopen), you're volunteering to associate your name to that vote (after all, your name is displayed once the question is actually closed/reopened), so I don't feel like anybody can reasonably object that they meant to keep their identity private when voting to close. The main reason I think it's good to not display who has voted to close a question which is not yet closed is that people vote differently when they know who is "on their side". Not displaying who has voted to close is a way of getting people to take personal responsibility for their vote to close. The dump contains so little information that undermines this purpose that I don't think it's a problem. I'm happy to change it back if somebody has a good reason to do so.

Soon after Anton reads this, you can access dumps at http://dumps.mathoverflow.net/. (Anton, do the usual! I'll give you shell access as well.)

http://dumps.mathoverflow.net/, or
http://ifile.it/soyqa09/MOdump20100303.zip