[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero reassigned HSEARCH-289: --------------------------------------- Assignee: Sanne Grinovero -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero reassigned HSEARCH-289: --------------------------------------- Assignee: Emmanuel Bernard (was: Sanne Grinovero) is this task meant only to take decisions? then it belongs to you. My opinion is we should use it. Also it doesn't break older implementations. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero commented on HSEARCH-289: ----------------------------------------- should I get this issue back and commit future updates using this JIRA code, or is this going to be closed and an "improvement" be opened? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero commented on HSEARCH-289: ----------------------------------------- no patch is needed to get the new interface working, the super (abstract) Filter does have a method providing backwards compatibility. The patch I'm working on will have no public effect, it's just about making better use of the same API, but I still need to test all performance effects and that could take some time (read 1-2 days). Should I close this? the patch is about an improvement, not a task about API evaluation, which I think is done at this point. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Emmanuel Bernard commented on HSEARCH-289: ------------------------------------------ An impl of chained filter using DocIdSet is available in Lucene misc for inspiration AFAIR I had to write my own but looked at this one during the BitSet times. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero commented on HSEARCH-289: ----------------------------------------- ah yes that's been a useful tip. what I did now is to build a DocIdBitSet from the "anded iterators", so that it is still possible to cache the expensive work. There are some ways it can be improved: as the client is seeing only an iterator, we could initialize the bitset partially ( up to the point the client is requesting, or from some point as the API is having a skipTo(int target) as well) but this sort of considerations I'll leave pending for when there is more time. another improvement would be to check if all chained sets are actually implemented throught a bitset, to cast and use the bitset and operations. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Emmanuel Bernard commented on HSEARCH-289: ------------------------------------------ Thanks Sanne! I applied your patch as I needed it for the book. I did a quick sanity check and it seems OK. Let get someone with proper hours of sleep check it before closing. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Sanne Grinovero commented on HSEARCH-289: ----------------------------------------- It improved just a little bit more; to get real performance we need to cast to DocIdBitSet when possible. Opening a new improvement for that. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-289?pag... ] Emmanuel Bernard commented on HSEARCH-289: ------------------------------------------ don't forget the OpenBitSet one too. You can apply the same kind of optimization -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira