It's looking very good; I'll seek a couple of hours this weekend to actually try it for real. I don't think it need changes, but if you seek for in depth-criticism I might add some additional fuel: I'm not 100% convinced about the use of "matching()" as method name, it confuses me a bit. This might be because of my non-native english, not sure. Doesn't "match()" sound better? "searchMatch" ? Having: monthQb .phrase() .slop( 1 ) .onField( "mythology" ) .sentence( "Month whitening" ) Here "sentence" takes the same logical role as "matching" ? WDYT about monthQb .phraseQuery() // stress that's a kind of Query .havingSlop( 1 ) || .withSlop( 1 ) .onField( "mythology" ) .searchSentence( "Month whitening" ) ? The method names are longer, but Imho there's a left part of the sentence and a right part of the sentence; the left part migh have options and be compound, the right part too. Waht is curretnyl missing to make it better readable is to choose appropriate nouns to have a clear separation to demark the end of the left part and the beginning of the right part. In the example above "slop" might look like a command, it sounds to me like "do a slop with option 1", of course that doesn't make sense, but is impairing readability. "searchSentence" "searchMatch" might be good delimiters for interpretation? Sanne 2010/6/3 Hardy Ferentschik <hibernate(a)ferentschik.de&gt;: > Looks fine to me looking at the examples. I haven't tried myself writing > my own queries though > to see the full potential. > > On Wed, 02 Jun 2010 18:00:20 +0200, Emmanuel Bernard > <emmanuel(a)hibernate.org&gt; wrote: > >> Guys, >> I'me now done with the level of abstraction and fluidity I wanted out of >> the query DSL. Please review before we push that out. Key features: >> - fluent API >> - use the field bridge system: type is passed, not raw string >> - use the analyzer transparently >> - simple use case simple, complex use cases possible >> >> I'm showing below a few examples demonstrating key concepts. >> >> Please comment / ask questions. >> >> Term query >> query = monthQb >> .keyword() >> .onField( "monthValue" ) >> .matching( 2 ) //note that monthValue is of type int >> .createQuery(); >> >> //term query, showing analyzer integration >> query = monthQb >> .keyword() >> .onField( "mythology_ngram" ) >> .matching( "snobored" ) //we apply the ngram filter here >> .createQuery(); >> >> //use fuzzy query >> query = monthQb >> .keyword() >> .fuzzy() >> .threshold( .8f ) //optional >> .prefixLength( 1 ) //optional >> .onField( "mythology" ) >> .matching( "calder" ) >> .createQuery(); >> >> //use wildcard queries >> monthQb >> .keyword() >> .wildcard() >> .onField( "mythology" ) >> .matching( "mon*" ) >> .createQuery(); >> >> Alternative option >> //apply on multiple fields >> monthQb.keyword() >> .onField( "mythology" ) >> .boostedTo( 30 ) >> .andField( "history" ) >> .matching( "whitening" ) >> .createQuery(); >> >> //boost a field >> monthQb >> .keyword() >> .onField( "history" ) >> .boostedTo( 30 ) >> .matching( "whitening" ) >> .createQuery(); >> >> Range query >> //Range query >> monthQb >> .range() >> .onField( "estimatedCreation" ) >> .from( from ) #from and to are actual java.util.Date. We do the >> conversion >> .to( to ).exclude() >> .createQuery(); >> >> monthQb >> .range() >> .onField( "estimatedCreation" ) >> .below( brithDay ) >> .createQuery(); >> >> Phrase query >> monthQb >> .phrase() >> .slop( 1 ) >> .onField( "mythology" ) >> .sentence( "Month whitening" ) >> .createQuery(); >> >> Boolean query >> monthQb >> .bool() >> .should( monthQb.keyword().onField( "mythology" ).matching( >> "whitening" ).createQuery() ) >> .should( monthQb.keyword().onField( "history" ).matching( "whitening" >> ).createQuery() ) >> .createQuery(); >> >> //Boolean query all except (recommended) >> monthQb >> .all() >> .except( monthQb.keyword().onField( "mythology" ).matching( "colder" >> ).createQuery() ) >> .createQuery(); >> >> _______________________________________________ >> hibernate-dev mailing list >> hibernate-dev(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/hibernate-dev > > _______________________________________________ > hibernate-dev mailing list > hibernate-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/hibernate-dev >

Please vote for: - 1. or 2. - matching vs whichMatches vs match - matchingSentence vs basedOnSentence - exclude vs excludeLimit - if 2. findInRange vs findWithinRange If you vote for 2., does the boolean query structural difference bother you? 1. monthQb .phrase() .withSlop( 1 ) .onField( "mythology" ) .sentence( "Month whitening" ); monthQb .keyword() .onField( "monthValue" ) .matching( 2 ) //.whichMatches ? .createQuery(); monthQb .keyword() .fuzzy() .onField( "mythology" ) .matching( "draggon" ) //.whichMatches ? .createQuery(); monthQb .range() .onField( "estimatedCreation" ) .from( from ) .to( to ).excludeLimit() .createQuery(); monthQb .bool() .should( qb.....createQuery() ) .should( qb.....createQuery() ) .createQuery(); ---------------------- 2. monthQb .findPhrase() .withSlop( 1 ) .onField( "mythology" ) .matchingSentence( "Month whitening" ) //basedOnSentence ? .includingWord("snow", 2);; monthQb .findkeyword() .onField( "monthValue" ) .matching( 2 ) .createQuery(); monthQb .findKeyword() .fuzzy() .onField( "mythology" ) .matching( "draggon" ) .createQuery(); monthQb .findInrange() // findWithinRange? .onField( "estimatedCreation" ) .from( from ) .to( to ). excludeLimit() .createQuery(); monthQb .bool() .should( qb.findKeyword().onField("mythology").matching("dragon").createQuery() ) .should( qb.....createQuery() ) .createQuery(); On 4 juin 2010, at 15:08, Emmanuel Bernard wrote: > withSlop instead of slop looks like a good improvement (same for the > fuzzy params etc) > > I have not used phraseQuery() because what you get is not a query > compared to createQuery(). That's why I think phrase is better than > phraseQuery. But I'm open to other options. > > matching vs sentence etc > There are definitely inconsistencies here :) > "keyword on field A matching B" > "range on field A from B to C" > > findKeyword, findInRange, findPhrase would be better? > what about boolean queries? > If e use this verb approach, "matching" is better than "match" > > "phrase on field A sentence" is wrong. I also want later in time to add > the ability to set words with their position. > phrase().onField("A").includingWord("C", 2); > phrase().onField("A").includingWord("C", 2); > What would be an alternative to "sentence" then? > > On 3 juin 2010, at 16:40, Sanne Grinovero wrote: > >> It's looking very good; I'll seek a couple of hours this weekend to >> actually try it for real. >> I don't think it need changes, but if you seek for in depth-criticism >> I might add some additional fuel: >> >> I'm not 100% convinced about the use of "matching()" as method name, >> it confuses me a bit. This might be because of my non-native english, >> not sure. >> Doesn't "match()" sound better? "searchMatch" ? >> >> Having: >> monthQb >> .phrase() >> .slop( 1 ) >> .onField( "mythology" ) >> .sentence( "Month whitening" ) >> >> Here "sentence" takes the same logical role as "matching" ? >> >> WDYT about >> >> monthQb >> .phraseQuery() // stress that's a kind of Query >> .havingSlop( 1 ) || .withSlop( 1 ) >> .onField( "mythology" ) >> .searchSentence( "Month whitening" ) >> ? >> The method names are longer, but Imho there's a left part of the >> sentence and a right part of the sentence; the left part migh have >> options and be compound, the right part too. Waht is curretnyl missing >> to make it better readable is to choose appropriate nouns to have a >> clear separation to demark the end of the left part and the beginning >> of the right part. >> In the example above "slop" might look like a command, it sounds to me >> like "do a slop with option 1", of course that doesn't make sense, but >> is impairing readability. >> >> "searchSentence" >> "searchMatch" >> might be good delimiters for interpretation? >> >> Sanne >> >> 2010/6/3 Hardy Ferentschik <hibernate(a)ferentschik.de&gt;: >>> Looks fine to me looking at the examples. I haven't tried myself >>> writing >>> my own queries though >>> to see the full potential. >>> >>> On Wed, 02 Jun 2010 18:00:20 +0200, Emmanuel Bernard >>> <emmanuel(a)hibernate.org&gt; wrote: >>> >>>> Guys, >>>> I'me now done with the level of abstraction and fluidity I wanted >>>> out of >>>> the query DSL. Please review before we push that out. Key features: >>>> - fluent API >>>> - use the field bridge system: type is passed, not raw string >>>> - use the analyzer transparently >>>> - simple use case simple, complex use cases possible >>>> >>>> I'm showing below a few examples demonstrating key concepts. >>>> >>>> Please comment / ask questions. >>>> >>>> Term query >>>> query = monthQb >>>> .keyword() >>>> .onField( "monthValue" ) >>>> .matching( 2 ) //note that monthValue is of type int >>>> .createQuery(); >>>> >>>> //term query, showing analyzer integration >>>> query = monthQb >>>> .keyword() >>>> .onField( "mythology_ngram" ) >>>> .matching( "snobored" ) //we apply the ngram filter here >>>> .createQuery(); >>>> >>>> //use fuzzy query >>>> query = monthQb >>>> .keyword() >>>> .fuzzy() >>>> .threshold( .8f ) //optional >>>> .prefixLength( 1 ) //optional >>>> .onField( "mythology" ) >>>> .matching( "calder" ) >>>> .createQuery(); >>>> >>>> //use wildcard queries >>>> monthQb >>>> .keyword() >>>> .wildcard() >>>> .onField( "mythology" ) >>>> .matching( "mon*" ) >>>> .createQuery(); >>>> >>>> Alternative option >>>> //apply on multiple fields >>>> monthQb.keyword() >>>> .onField( "mythology" ) >>>> .boostedTo( 30 ) >>>> .andField( "history" ) >>>> .matching( "whitening" ) >>>> .createQuery(); >>>> >>>> //boost a field >>>> monthQb >>>> .keyword() >>>> .onField( "history" ) >>>> .boostedTo( 30 ) >>>> .matching( "whitening" ) >>>> .createQuery(); >>>> >>>> Range query >>>> //Range query >>>> monthQb >>>> .range() >>>> .onField( "estimatedCreation" ) >>>> .from( from ) #from and to are actual java.util.Date. >>>> We do the >>>> conversion >>>> .to( to ).exclude() >>>> .createQuery(); >>>> >>>> monthQb >>>> .range() >>>> .onField( "estimatedCreation" ) >>>> .below( brithDay ) >>>> .createQuery(); >>>> >>>> Phrase query >>>> monthQb >>>> .phrase() >>>> .slop( 1 ) >>>> .onField( "mythology" ) >>>> .sentence( "Month whitening" ) >>>> .createQuery(); >>>> >>>> Boolean query >>>> monthQb >>>> .bool() >>>> .should( monthQb.keyword().onField( "mythology" >>>> ).matching( >>>> "whitening" ).createQuery() ) >>>> .should( monthQb.keyword().onField( "history" >>>> ).matching( "whitening" >>>> ).createQuery() ) >>>> .createQuery(); >>>> >>>> //Boolean query all except (recommended) >>>> monthQb >>>> .all() >>>> .except( monthQb.keyword().onField( "mythology" >>>> ).matching( "colder" >>>> ).createQuery() ) >>>> .createQuery(); >>>> >>>> _______________________________________________ >>>> hibernate-dev mailing list >>>> hibernate-dev(a)lists.jboss.org >>>> https://lists.jboss.org/mailman/listinfo/hibernate-dev >>> >>> _______________________________________________ >>> hibernate-dev mailing list >>> hibernate-dev(a)lists.jboss.org >>> https://lists.jboss.org/mailman/listinfo/hibernate-dev >>> > > > _______________________________________________ > hibernate-dev mailing list > hibernate-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/hibernate-dev _______________________________________________ hibernate-dev mailing list hibernate-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/hibernate-dev

> Even though I understand the conceptional difference between 'matching' > and 'sentence' when it comes to a phrase > query, but is there really a need for two different methods? From the > context I can always see whether I have a > keyword() or phrase() query. I've been thinking along your lines for a while. I wanted to make it explicit that we were passing a full sentence and not key words but that's not a strong choice. That would leave matchingSentence vs basedOnSentence vs matching