Hmm - yeah - coming at that a bit from the Lucene perspective - you wouldn't want it cached. For a ton of clauses like that, a filter can outperform a booleanquery on a large index in certain cases (an uncached filter). And doesn't suffer from having to mess with the max clause setting. Thats part of why the Multi-Term queries now use a constant score mode (essentially a filter) when matching a lot of terms. They use a constant score boolean query mode when matching fewer terms.
Giovanni Fernandez-Kincade wrote: > Is a filter really appropriate in this case? These types of condition would > never get re-used - they would be unique to each search performed by users in > our application. > > Is there any performance benefit to the filter aside from caching (which > wouldn't do much for us anyway)? > > > > -----Original Message----- > From: Mark Miller [mailto:markrmil...@gmail.com] > Sent: Thursday, October 22, 2009 6:31 PM > To: solr-user@lucene.apache.org > Subject: Re: Too Many Boolean Clauses > > Giovanni Fernandez-Kincade wrote: > >> Hi, >> I'm trying to perform a search against an integer field with a ton of OR >> statements for each of the unique values that I want to search for. I pasted >> an example at the bottom of this email. Solr fires back the following error: >> org.apache.lucene.queryParser.ParseException: Cannot parse .. ': too many >> boolean clauses >> Can you configure the maximum number of boolean clauses in lucene? Is there >> a better way to perform this type of search? I can't use numeric range >> queries because we often need to search for a huge number of IDs that are >> non-continuous (e.g. 4, 64, 1003, 2490, etc.). I feel like anyone that is >> combining SOLR with a relational database and has Solr documents referencing >> DB entities must have run into this issue. >> >> Thanks, >> Gio. >> >> Field being searched: >> <field name="formTypeId" type="integer" indexed="true" stored="false" >> multiValued="true"/> >> >> Example query: >> +companyId:32393 +formTypeId:(5 OR 6 OR 59 OR 125 OR 126 OR 135 OR 136 OR >> 197 OR 198 OR 209 OR 210 OR 211 OR 212 OR 213 OR 214 OR 230 OR 231 OR 232 OR >> 233 OR 298 OR 299 OR 377 OR 378 OR 394 OR 395 OR 402 OR 428 OR 429 OR 659 OR >> 670 OR 671 OR 825 OR 894 OR 3245 OR 3246 OR 3247 OR 3248 OR 3249 OR 3250 OR >> 3251 OR 3252 OR 3253 OR 3254 OR 3255 OR 3256 OR 3257 OR 3258 OR 3259 OR 3260 >> OR 3261 OR 3262 OR 3263 OR 3264 OR 3265 OR 3266 OR 3267 OR 3268 OR 3269 OR >> 3270 OR 3275 OR 3276 OR 3277 OR 3280 OR 3281 OR 3282 OR 3283 OR 3284 OR 3285 >> OR 3286 OR 3287 OR 3288 OR 3289 OR 3290 OR 3291 OR 3292 OR 3293 OR 3294 OR >> 3295 OR 3296 OR 3297 OR 3298 OR 3299 OR 3300 OR 3301 OR 3302 OR 3303 OR 3304 >> OR 3305 OR 3306 OR 3307 OR 3308 OR 3309 OR 3310 OR 3311 OR 3312 OR 3313 OR >> 3314 OR 3315 OR 3316 OR 3317 OR 3318 OR 3319 OR 3320 OR 3321 OR 3322 OR 3323 >> OR 3324 OR 3325 OR 3892 OR 3893 OR 3894 OR 3895 OR 3896 OR 3897 OR 3900 OR >> 3901 OR 3903 OR 3904 OR 3905 OR 3906 OR 3907 OR 3908 OR 3909 OR 3910 OR 3911 >> OR 3912 OR 3913 OR 5012 OR 5013 OR 5014 OR 5015 OR 5016 OR 5019 OR 5020 OR >> 5021 OR 5022 OR 5023 OR 5024 OR 5025 OR 5026 OR 5027 OR 5028 OR 5029 OR 5030 >> OR 6069 OR 6070 OR 6084 OR 6085 OR ... >> >> >> > Yuck. Might want to try making a filter instead. Or something else. > > You can raise the maxBooleanClause limit in solrconfig.xml. You can find > an example in the example config. > > -- - Mark http://www.lucidimagination.com
