From: Mark T. <mt...@nl...> - 2008-03-31 20:25:58
|
Hey Andrew, Andrew Nagy <and...@vi...> writes: >> I'm not sure what the current state of support is, but VuFind 0.6 >> didn't *quite* work out of the box. Solr handles unicode characters >> with no problems (so searching works), but we had to make some changes >> to the PHP to escape multi-byte characters properly, some changes to >> the record display code to pull out the contents of the 880 fields, >> and some changes to the importer to index them. > > Mark - any chance you guys will be returning your modifications? I > would be more than happy to add you guys as committers to the vufind > SVN. We're still keen to return what we can, although slightly scared by the fact that our code bases have diverged quite a lot since 0.6. We've been holding off raising it because we've both been developing to heavily and merging two moving targets is hard :o) With regard to unicode characters, we're not quite finished defining our Solr indexes ourselves, as pulling the data out of these 880 fields is a little bit fiddly. I could send a patch for the basic PHP character handling stuff, though, as this will at least stop PHP from munging Unicod characters in search strings. > Also - the librarians on staff here have been playing with your site and > say you're searching works better - have you done some customizations to > the search queries? Yeah, it's probably the area we've done the most customisation. Searches are now parsed and manipulated in PHP land, and what we end up sending to Solr is quite a different beast to the user's original query. We aimed to do this all in a fairly generic way that we could share our changes, so I guess this is the test :o) I'm not sure I'd want to dive straight into SVN and start making these sort of big changes, but what if I check out your latest SVN head and create a version with our search code merged in? At least that way I could send you a patch and let you decide if you're interested in using our changes. I'll see what I can do over the next couple of days. Cheers, Mark -- Mark Triggs <mt...@nl...> |