The short answer is that unique keys should be s single term. String types are guaranteed to be single, since they aren't analyzed. Your SplitUpStuff type *does* analyze terms, and can make multiple tokens out of single strings via WordDelimterFactory.
A common error when thinking about the "string" the type is not understanding that it is NOT analyzed. It's indexed as a single term. So whey you define UniqueKey of type string, it behaves as you expect. That is documents are updated if the ID field matches exactly, case, spaces, order and all. By introducing your "SplitUpStuff" type as UniqueKey, Well, I don't even know what behavior I'd expect. And whatever behavior I happened to observe would not be guaranteed to be the behavior of the next release. Consider what you're asking for and you can see why you don't want to analyze your uniquekey field. Consider the following simple text type (where each word is a term). You have two values from two different docs doc1: "this is a nice unique key" doc2: "My Keys are Unique and Nice" It's quite possible, with combinations of analyzers and stemmers to index the exact same tokens, namely "nice", "unique" and "key" for each document. Are these equivalent? Does order count? Capitalization? It'd just be a nightmare to try to explain/predict/implement. Likely whatever behavior you do get is just whatever falls out of the code. I'm not even sure any attempt is made to enforce uniqueness on an analyzed field. HTH Erick On Sun, Aug 15, 2010 at 11:59 AM, j <jta...@gmail.com> wrote: > I guess another way to pose the question is- what could cause > <uniqueKey>id</uniqueKey> to no longer be respected? > > > The last chance I made since I noticed the problem of non-unique docs > was by changing field "title" from "string" to "SplitUpStuff". But I > dont understand how that could affect the uniqueness of a different > field called "id". > > <fieldType name="splitUpStuff" class="solr.TextField" > positionIncrementGap="100"> > <analyzer type="index"> > <filter class="solr.WordDelimiterFilterFactory" > generateWordParts="1" generateNumberParts="0" c > <filter class="solr.StopFilterFactory" > ignoreCase="true" > words="stopwords.txt" > enablePositionIncrements="false" > /> > <filter class="solr.LowerCaseFilterFactory"/> > <filter class="solr.EnglishPorterFilterFactory" > protected="protwords.txt"/> > <filter class="solr.RemoveDuplicatesTokenFilterFactory"/> > </analyzer> > </fieldType> > > > > > > > In order to make even a guess, we'd have to see your new > field type. Particularly its field definitions and the analysis > chain... > > Best > Erick > > On Fri, Aug 13, 2010 at 5:16 PM, j <jta...@gmail.com> wrote: > > > Does fieldType have any effect on the thing that I specify should be > > unique? > > > > uniqueKey has been working for me up until recently. I change the > > field that is unique from type "string" to a fieldType that I have > > defined. Now when I do an update I get a newly created document (so > > that I have duplicates). > > > > Has anyone else had this problem before? > > >
