[
https://issues.apache.org/jira/browse/SOLR-5057?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13718377#comment-13718377
]
Yonik Seeley commented on SOLR-5057:
------------------------------------
I think the common case will be fqs being ordered (most fqs will be generated
from a system and hence normally appear in the same order.)
A few issues with this patch:
- the sort is expensive - it generates hashcodes for elements on every single
compare.
- it's still not 100% - hashcodes for two different queries could match so
sorting doesn't guarantee to get in the right order
- it slows down the common case of ordered filters (the sort needs to be done
even when just checking the cache too)
How about this (untested) code instead:
{code}
// Do fast version, expecting that filters are ordered and only
// fall back to unordered compare on the first non-equal elements.
// This will only be called if the hash code of the entire key already
// matched, so the slower unorderedCompare should pretty much never
// be called if filter lists are generally ordered.
private static boolean isEqual(List<Query> a, List<Query> b) {
if (a==b) return true; // takes care of identity and null cases
if (a==null || b==null) return false;
int sz = a.size();
if (sz != b.size()) return false;
for (int i=0; i<sz; i++) {
if (!a.get(i).equals(b.get(i))) {
return unorderedCompare(a, b, i);
}
}
return true;
}
private static boolean unorderedCompare(List<Query> a, List<Query> b, int
start) {
int sz = a.size();
outer: for (int aIdx = start; aIdx<sz; aIdx++) {
Query aQ = a.get(aIdx);
for (int bIdx = start; bIdx<sz; bIdx++) {
Query bQ =b.get(bIdx);
if (aQ.equals(bQ)) continue outer;
}
return false;
}
return true;
}
{code}
> queryResultCache should not related with the order of fq's list
> ---------------------------------------------------------------
>
> Key: SOLR-5057
> URL: https://issues.apache.org/jira/browse/SOLR-5057
> Project: Solr
> Issue Type: Improvement
> Components: search
> Affects Versions: 4.0, 4.1, 4.2, 4.3
> Reporter: Feihong Huang
> Assignee: Erick Erickson
> Priority: Minor
> Attachments: SOLR-5057.patch, SOLR-5057.patch
>
> Original Estimate: 48h
> Remaining Estimate: 48h
>
> There are two case query with the same meaning below. But the case2 can't use
> the queryResultCache when case1 is executed.
> case1: q=*:*&fq=field1:value1&fq=field2:value2
> case2: q=*:*&fq=field2:value2&fq=field1:value1
> I think queryResultCache should not be related with the order of fq's list.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
