See LR. I plan to eventually make this page have the content and the LR page to point here. I'm increasingly beginning to feel that wikis need to be more intelligent and clean up this kind of mess for you.
Interesting idea! How do you envision it working, from an end-user perspective? Maybe a 'rename' button would do; if we also had page forwarding, it could create a new page with the same content, and set up a redirect from the old page. A backlink search would also indicate where links to the old page could be found, and these could be changed to point to the new page; this might be a bit heavy-handed, though, and not always appropriate.
Well originally this wish was for AI to go and clean things up, but on a more serious note, I think it boils down to: page aliasing/redirection, and a CVS attic for "deleting" things. Old links should still work in case 1, and say "not found but you can find it here" for 2 (and have an autogen page of broken links).
The redirection is on the to-do list, but i haven't quite figured out how to do it. I feel it should be effected by writing something in the page's text, with my best idea so far being that if the first block begins with a '>', it's a redirection instruction. However, it is then not clear how one would edit a page which redirects, since you could't ever be on a view page for it which would offer you an edit link. Hmm. Perhaps the instruction should not be for redirection, but transclusion; that would have the effect of getting the wrong name to the right information without duplication, and might even be useful in other situations, whilst keeping a view page around in case you want to edit it. On the other hand, it might be abused, with transclusion of all sorts of stuff all over the place. Also, we need to protect against infinite loops in any redirection or transclusion system. Disappointingly but usefully, i find that the idea of Wiki Transclusion is not new.
A CVS-like Wiki History is also on the cards, but further off. It probably won't use CVS, as that's a bit heavyweight, but it will use a CVS-like scheme of serial diffs.
With the system i envision, pages could not both exist and be a redirection point at the same time. There would thus be a choice about what to do with redundant pages; evacuate the content, fix the inbound links, and then either delete them, or leave them as a redirection. The former is cleaner, but the latter helps if someone writes a link to the page in the future (which may be likely for things like common misspellings etc).
I do have a script which finds Dangling Wikinames, but at present it only runs from the command line. It is a definite for inclusion in the wiki proper, but not before the new Back Links Implementation is in place (which in turn will only come with the Great Twic I Rewrite).
I think that's why page aliasing is nicer. By having the first line read, say "AKA Foo,Bar", you can edit the page and the aliases start/stop happening. It does mean that on page save, the wiki needs to check and add it to some sort of database (although dbm would do nicely).
I'm due to find all of this out now I have a home brewed wiki effort ( albeit content-free - http://public.sys3175.co.uk )
|Tue, 08 Jul 2003 14:11:48 GMT||Front Page||Recent Changes||Message Of The Day|