[P.S. to my first post] Further contemplating http://wiki.apache.org/solr/FunctionQuery.
I am using 1.4.1, the date field is configured like this: <fieldType name="date" class="solr.DateField" sortMissingLast="true" omitNorms="true"/> (The schema has been created using the schema file from 1.4.0, and I haven't changed anything when upgrading to 1.4.1. TrieDate is said to be the default in 1.4, so I would expect this date field to have that type?) On the wiki page, the following example is listed: Example: ms(NOW/DAY) Could I do that same thing with my own date? ms(start_date/DAY) I tried that query: http://192.168.2.40:8080/solr/epg/select?qt=dismax&fl=start_date,title&sort=ms%28start_date/DAY%29%20asc,title%20asc (search for all *:* configured in solrconfig.xml for dismax) I get the following error message back: """ message can not sort on undefined field: ms(start_date/DAY) description The request sent by the client was syntactically incorrect (can not sort on undefined field: ms(start_date/DAY)). """ I am a complete newbie when it comes to function queries. Thanks for any suggestions! Chantal On Fri, 2010-07-09 at 11:44 +0200, Chantal Ackermann wrote: > Dear all, > > this is not a new problem, I just wanted to check whether with 1.4 there > might have been changes that allow a different approach. > > In my query, I retrieve results that have a date field. I have to sort > the result by day only, then by a different string field. The time of > that date shall not be used for sorting. > I cannot filter the results on a certain date (day). > > This thread confirms my first thought that I need another field in the > index: > http://search.lucidimagination.com/search/document/422dc30e0a222c28/sorting_dates_with_reduced_precision#46566037750d7b5 > > However, is it possible to use the DateMathParser somehow in the > function queries? > If it's not yet possible - why not: > (a) is there are great risk that the performance would be bad? Or some > other reason that discourages this solution. > (b) simple not implemented > > In case of (b), I might try to implement it. > > Thanks! > Chantal >
