Last modified: 2014-02-20 21:21:39 UTC
Compare the following: Barack Hussein Obama II. (Barack-Hussein-Obama-en-US-pronunciation. ogg | b | ə | ˈ | r | ɑː | k | _ | h | uː | ˈ | s | eɪ | n | _ | oʊ | ˈ | ... - vs. - Barack Hussein Obama II is the 44th and current President of the United States, having taken office in 2009. He is the first African American... or The Kingdom of Saudi Arabia. (المملكة العربية السعودية ar | Al Mamlaka al ʻArabiyya as Suʻūdiyya commonly known as Saudi Arabia. us-Saudi Arabia-... - vs. - The Kingdom of Saudi Arabia is, in land area, the third largest Arab country and the largest country in the Middle East. It is bordered by... The first versions are from our current search results. The second versions are what the search results would look like if we skipped the part in parentheses (like Google seems to do in most cases).
While we're at it, we should probably remove ref tags as well.
I was going to try to patch this in on truck, but search on my trunk checkout seems to be totally wonky. In particular, it doesn't seem to give beginning snippets for title matches, but rather term-highlighting snippets from the middle of the article (which often ends up just being the interlanguage links). If we're not doing beginning snippets any more, this bug should be marked invalid.
OK, it looks like we have 3 different pieces of software for doing search, and my local install is not set up the same way that the search on the cluster is. It's likely that this bug isn't even filed under the right product and component. If anyone knows more about search, feel free to move it.
Looks like this probably belongs under the Lucene-search extension.
Which, feel free to look at that and take a stab at it if you have any interest in Java. I don't think we have an active maintainer for that.
[Merging "MediaWiki extensions/Lucene Search" into "Wikimedia/lucene-search2", see bug 46542. You can filter bugmail for: search-component-merge-20130326 ]
Moving to CirrusSearch.
