Upgrade to Lucene Core 3.1?


(aditya.kulkarni) #1

Hey,
Is update to Lucene Core 3.1 scheduled anytime soon?
Best Regards,
aditya


(Shay Banon) #2

Yes, it already in master. Is there a feature that you need from it?
On Monday, April 11, 2011 at 4:57 PM, aditya.kulkarni wrote:

Hey,
Is update to Lucene Core 3.1 scheduled anytime soon?
Best Regards,
aditya


Best Regards,
a

View this message in context: http://elasticsearch-users.115913.n3.nabble.com/upgrade-to-Lucene-Core-3-1-tp2806662p2806662.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(aditya.kulkarni) #3

Hey Shay,

Yes, that is correct.

  1. There are some modifications to IndexWriter for working with HBase as long term storage (non-indexed data, but required for running jobs). We based it along with file copy method from stable branch for quite sometime (quite sometime spent on that now :), so don't want to throw it away)
  2. also finding new functionality of TotalHitCountCollector useful on standalone tests
  3. inherent optimizations that comes with new release (still exploring)

It's good that ES master has it (but can't get to compile it; gradle test keeps on failing; failing occurs at different places on linux and windows platform respectively)

Master refresh for lucene I trust will come with 16.0, right?

-a


(Shay Banon) #4

How do the first two points relate to elasticsearch in any way?
On Monday, April 11, 2011 at 5:23 PM, aditya.kulkarni wrote:

Hey Shay,

Yes, that is correct.

  1. There are some modifications to IndexWriter for working with HBase as
    long term storage (non-indexed data, but required for running jobs). We
    based it along with file copy method from stable branch for quite sometime
    (quite sometime spent on that now :), so don't want to throw it away)
  2. also finding new functionality of TotalHitCountCollector useful on
    standalone tests
  3. inherent optimizations that comes with new release (still exploring)

It's good that ES master has it (but can't get to compile it; gradle test
keeps on failing; failing occurs at different places on linux and windows
platform respectively)

Master refresh for lucene I trust will come with 16.0, right?

-a


Best Regards,
a

View this message in context: http://elasticsearch-users.115913.n3.nabble.com/upgrade-to-Lucene-Core-3-1-tp2806662p2806775.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(Lukáš Vlček) #5

Hi,

master compiles without issues for me with Lucene 3.1 do you have more
details? As far as I know running "gradlew clean release" does not include
"test" by default, so failing tests should not impact possibility to produce
release (the fact that tests can fail when not started from IDEA is probably
a different story).

Lukas

On Mon, Apr 11, 2011 at 4:23 PM, aditya.kulkarni
aditya.kulkarni@gmail.comwrote:

Hey Shay,

Yes, that is correct.

  1. There are some modifications to IndexWriter for working with HBase as
    long term storage (non-indexed data, but required for running jobs). We
    based it along with file copy method from stable branch for quite sometime
    (quite sometime spent on that now :), so don't want to throw it away)
  2. also finding new functionality of TotalHitCountCollector useful on
    standalone tests
  3. inherent optimizations that comes with new release (still exploring)

It's good that ES master has it (but can't get to compile it; gradle test
keeps on failing; failing occurs at different places on linux and windows
platform respectively)

Master refresh for lucene I trust will come with 16.0, right?

-a


Best Regards,
a

View this message in context:
http://elasticsearch-users.115913.n3.nabble.com/upgrade-to-Lucene-Core-3-1-tp2806662p2806775.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(aditya.kulkarni) #6

Hi Shay:

We are interested in having data searchable immediately as it enters system. Hence, for now what we are doing is that we are making ES get input data set with specified analyze/do not analyze. Fields those are analyzed and indexed are immediately part of search experience.
Based on where as backend, we are overriding indexwriter to simultaneously give conditional dump to Hbase for storage and jobs on top of it.

below is architecture that we are trying to use. We are using ES in context of complete layer which allows free flow schema generation plus is gives search.

Data
|
[ES]
|
[modified Lucene index writer]
| |
[ES storage] [HBase]

Lukáš: I will respond to your question separately. Starting fresh again. machine - Ubuntu 32bit.

Best Regards,
Aditya


(aditya.kulkarni) #7

Hi Lukas,

Got your point & it's correct. Build is happening. It's gradle test that is failing. tomorrow I will continue with testing that it works perfectly fine.

Best Regards,
Aditya


(Shay Banon) #8

Run gradle release, it will create the distribution.
On Monday, April 11, 2011 at 6:21 PM, aditya.kulkarni wrote:

Hi Lukas,

Got your point & it's correct. Build is happening. It's gradle test that is
failing. tomorrow I will continue with testing that it works perfectly fine.

Best Regards,
Aditya


Best Regards,
a

View this message in context: http://elasticsearch-users.115913.n3.nabble.com/upgrade-to-Lucene-Core-3-1-tp2806662p2807003.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(system) #9