User blog comment:Rappy 4187/Technical Update: November 2, 2015/@comment-11733175-20151104095739/@comment-11733175-20151108235240

@Kirk Much appreciated :)

@Deadcoder I'm not going to argue why one prefix is better than another - it's besides the point. The point is that a select number of extensions/features expect predefined prefixes. The two examples I've listed have been in use on Wikia for years, and are in use on a number of wikis although I would imagine it's limited to large wikis with noticeable vandalism (AbuseFilter) and wikis with users with more than average experience, compared to the average wikia user, of writing code. It's certainly possible that there are wikis that do not have users who both understand and are capable of utilising these extensions but have them anyway, particularly in the case of AF as I understand VSTF use it in certain cases.

As far as migrating the messages to a new namespace goes, your idea is about right. Gadgets is slowly being migrated to new namespaces in the current version of the extension, in fact the commit for adding the namespaces is actually in code review now. Tags could feasibly undergo the same change, but that would require a radically different approach especially to include the current flexibility with i18n. However, any changes in the 2 examples I've listed are extremely unlikely to ever arrive on Wikia due to their lack of desire to upgrade MediaWiki. So we're back to the current process of reviewing message usage.

What the new custom-* prefix does achieve is a way for users to continue to use customised messages, potentially adding functionality back to scripts such as Standard Edit Summaries, if the definitions were stored in MediaWiki:, and allowing the continued use of the int magic word (which is very rarely used in my experience). Another use case is preload templates as Kirkburn mentioned somewhere below. It's actually a fairly limited set of use cases, so I would guess we're actually nearing the end of message usage reviews and this is aimed at catching one of the few remaining use cases.