jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Thomas Mueller <muel...@adobe.com>
Subject Re: [jr3 optional features]
Date Fri, 24 Feb 2012 10:24:03 GMT
Hi,

>>The path isn't stored in Lucene even for Jackrabbit 2, only the node id.
>
>No, but the hierarchical relations *are* evaluated through Lucene in
>Jackrabbit 2

I don't fully know how the current index mechanism works so I can't
comment here.

>I don't think you can write evaluate these queries more efficient
>without Lucene (Slide used to do this back in the days). But at the
>same time, they are too complex to map to an inversed index.

So what your are saying is that no matter if you use Lucene or not it's
not possible? I don't agree.

The plan is to let users define indexes manually, and the query engine
(built by us) decides which index(es) to use; possibly we will add a
mechanism to let the user decide which indexes to use (hints). Just like
in databases. This will shift some of the burden to the user who is
supposed to define the right indexes. We will need quite a bit of
documentation how to do this. Even without the (right) indexes it will
work, just much slower.

>Either way, I am of course fine with any attempt, but I am just afraid
>that it will take lots of efforts to build, and in the end doesn't
>scale well enough (at least not to the targets that jr3 aims for).

I'm sure it will work quite well. It will be some effort, but compared to
query processing in SQL databases it's a lot simpler. I expect it to work
reasonably well at the end of this year.

Regards,
Thomas


Mime
View raw message