"Andrey A. Chernov" wrote: > > Rating always == 1 bug still present in 3.1.10 (as in 3.1.9 too). > > Now I use single text file contains 4 equal words to track it down. > In UdmFindSQL() I got correct result > wrd[0].pos == 4 > (which becomes "rating" later in Res->Doc[i].rating=wrd[i+first].pos) > but this value is _damaged_ (reset to 1) right after wrd[0].pos is word position when "Phrase yes" and the number of this word appearance in the document when "Phrase no" > UdmGroupByURL(query,wrd); > call. If I comment it out, I got correct rating. UdmGroupByURL() is > complex function, and I don't try to debug it, but this particular > statement is very suspicious (since it overwrites .pos): > wrd[j].pos=bit_count(wrd[j].count); > > Please send me fix or explaination, what is wrong here. Everything OK here."rating" means "number of unique query words found in a document". If you search for one word, it always 1. If you search for two word and both of them are in document, it will be 2, in the case when only one word presents rating will be 1. Displayed rating does not depend on count of each word in documents. However document ordering take it in account. ___________________________________________ If you want to unsubscribe send "unsubscribe general" to [EMAIL PROTECTED]
