Last modified: 2010-05-15 15:38:16 UTC
Expected: Link table updates should be run on pages created via Special:Import. Actual: Nope!
*** Bug 3078 has been marked as a duplicate of this bug. ***
// !FILE : SpecialImport.php function importOldRevision() just before returning true, add : $article->editUpdates($article->preSaveTransform($this->getText())) ;
I am not really sure the above 'preSaveTransform' stuff is really needed in this case, but, it will not harm...
In fact, my solution (above) solves the problem that imported articles were non 'searchable' but, until now, I did not find the way to automatically update a so-called 'indexing' within categories. So, my solution is probably worth to be implemented as it solves half of the problem. But there is still job to realize the other steps : - update categories - and, but this is much less important, updating the list of recent changes, etc. This is a really serious problem because, if we want to massively import data, we need to make them available thru categories. (In my case, I convinced ex-Windows victims to provide me with a stupid MS Access file including a few thousands of books descriptions, as to put them on a wiki, with categories, keywords, images, comments, etc... it works really fine... but they would also like to retrieve their books without the need to edit each article manually :-) ...).
I once suggested, in a different context (bug 939, comment 3) adding a flag in the database that causes a page's links to be re-processed the next time it was rendered - in other words, next time it was viewed, since the cache would also be purged at the same time. For Special:Import would probably make more sense to just run an "updateLinks()" function (which we don't really have yet, AFAIK) at the time of import, but I just thought I'd connect the thoughts.
I solved my personal problem : 1) by patching SpecialImport.php as explained in comment 2 (this immediately solves the 'search' part of the problem, so that we can see if and what was really uploaded) 2) by running the 'touch' (also named sometimes 'touchall') standard bot. (this solves the remainder parts of the problem)) However, it would be nice not to suggest the use of bot's. a) because not everybody is able to run a bot b) because, if everybody will start using bots, this can be very dangerous (for the contents and for the resources).
*** Bug 3320 has been marked as a duplicate of this bug. ***
*** Bug 4049 has been marked as a duplicate of this bug. ***
*** Bug 5769 has been marked as a duplicate of this bug. ***
Fixed in r15096.