Last modified: 2013-10-05 01:14:28 UTC
On www.mediawiki.org, I have over fifty imported edits on the page [[mw:User:Happy-melon/Templates]], however, these do not appear in any integrated editcount: the counter in Preferences, and the API ?meta=userinfo&uiprop=editcount and ?list=usercontribs&ucuser=Happy-melon all record only the six edits I actually made directly to the site. I can see how there could be licensing issues if the usernames represented distinct users on the two wikis, but given that I *am* the same user (both accounts are part of the my SUL and were at the time of the import), shouldn't these numbers account for the imported edits? Regardless, surely the situation whereby the edits *look* like they're from me in histories, contribs, diffs, etc, but aren't recorded as being mine, isn't right?
To clarify, the page was transwikied from meta to mediawiki. http://www.mediawiki.org/w/index.php?title=Special:Log&page=User%3AHappy-melon%2FTemplates I think the mistake is actually that the imported edits are attributed to your mediawiki username at all. See https://bugzilla.wikimedia.org/show_bug.cgi?id=18163 "Importing a page shouldn't attribute the edits to local usernames".
In general I agree with the sentiments of that bug and bug 13798 which it is duped to; *however* when both wikis are part of the SUL network that is not so sensible, as "happy-melon@enwiki" is the *same user* as "happy-melon" on mediawikiwiki. I guess the default implementation should be to attribute the edits to the user account on the old site, and add a hook that CentralAuth uses to instead attribute the edits to the local account if it exists.
(In reply to comment #0) > Regardless, surely the situation whereby the edits > *look* like they're from me in histories, contribs, diffs, etc, but aren't > recorded as being mine, isn't right? Special:Contributions was "fixed" because it looks now (also?) for user_id, which for imported edits is 0, so the situation you describe is no longer there. Anyway, unless the user_id is matched to the local user nothing can or should be done: it would be wrong to consider in any way edits we don't know the author of. If that was solved, the only left fragment of this bug would be the need to update some editcounts for the newly matching user_id's, but I wouldn't call that a bug. Closing.