It's not an option for my current problems, where I should integrate
already-partially-existing Lucene 4 code (especially using the facet and
spatial modules) into a Jackrabbit-based product.
In a first approach I thought migrating Jackrabbit to Lucene 4. I've
even thought about integrating faceting directly in Jackrabbit via the
Lucene facet module.
But since the migration is not an option neither, I'll dive into
ClassLoader isolation. It's quite funny as well :)
Regards,
Cédric
Le 09/10/2013 12:20, Torgeir Veimo a écrit :
How about using an external elasticsearch or solr instance instead of
lucene for your indexing needs?
On 9 October 2013 18:21, Cédric Damioli <[email protected]> wrote:
Hi,
Ok I see. Thanks for your feedback.
I most likely won't dive alone in this task, given that there's some really
tricky parts.
On the other hand, Oak is currently not production-ready, so I have to stick
to JR 2.x for now.
I'll probably look at a way to isolate Jackrabbit+Lucene in a separate
ClassLoader to be able to run Lucene 4 code in the same application. BTW,
did someone already achieve this (apart from OSGI-based solutions) ?
Regards,
Cédric
Le 09/10/2013 09:07, Marcel Reutegger a écrit :
Hi Cédric,
most of the development effort currently goes into Apache
Oak, which will replace Jackrabbit 2.x at some point. This means
Jackrabbit 2.x is pretty much in maintenance mode.
I don't want to discourage you from doing this, but I will probably
not have the time to help on this...
Regards
Marcel
-----Original Message-----
From: Cédric Damioli [mailto:[email protected]]
Sent: Montag, 7. Oktober 2013 17:59
To: [email protected]
Subject: Upgrade to Lucene 4 ?
Hi,
I'm thinking about the potential migration of Jackrabbit to Lucene 4.
This represents quite a huge work, as many Lucene core API have
completely changed between 3.x and 4.x
A quick test by replacing Lucene 3.x by Lucene 4.x shows almost 1000
compilation problems.
So before beginning this, I was first wondering if it is even worth
doing it.
The main benefits are obviously Lucene 4 new features and performance
optimizations.
The main drawback is IMHO that for a given application, a Jackrabbit
migration would also involve the migration of all Lucene-related code if
any. Existing repositories should also be reindexed during migration.
So what do people think about this ?
And of course, if it seems a good idea, is there some committers around
here volunteering to help me to do the job ? ;)
Best regards,
--
Cédric Damioli
CMS - Java - Open Source
www.ametys.org
