The Joy wrote:I'm confused. Are they giving up on Single-User-Logins and going back to requiring the old way of having to register for every Wikimedia site?
Nope. What I'm saying is, the table in the database that contains page revisions predates all that. Each entry, in addition to having the timestamp, content, etc., has the user's name
and internal ID in it. This means it's not in
second normal form. It should only have the user's ID, and the user name should live in another table, with the ID as the key. So the case now is that due to ancient fuckups a decade ago, some revisions have mismatched user names and user IDs associated with them, or even no ID at all. And the IDs aren't visible to end users, just the user names - those are what critical tools like Special:Contributions rely upon. I keep identifying these messes and having to ask people with database access to correct them, rather than the database having not been designed by a buffoon and thus preventing them from being possible in the first place.
Plus, what my earlier post linking to that bug report was referring to is that there's a tool which lets you clone revisions from one wiki to another. But it was built before the rationalization. Its "solution" to avoiding the possibility of edits by a remote user being attributed to a local user with the same name (when that situation was still possible) was to set the user ID to 0 while retaining the user name. Meaning that the revision would display a user name, but not actually be tied to a local account by that name. In other words, a shitty kludge. And somehow, in all this miserably drawn-out period of work to unify accounts, that was never addressed. Every time someone uses it, it creates revisions that don't contain the IDs of the marvellous new unique accounts that are such a great WMF Engineering achievement. The whole system is just a rabbit hole of stupid.