Hello,
I am planning to build a bigger Internet platform and actually
evaluating
Java EE and Rails. I have a lot of Java experience and I am quite new to
Rails. After playing some weeks with Rails I am sure that it is a mature
web
framework and I really like the productivtity of Rails. One of the key
advantages is that new developers will understand this platform much
quicker
than all the Java EE Apis and Frameworks.
Despite many features of our platform that will be easy to handle with
Rails, we need a good full text search on some Text database entries and
a
CMS for the content around the application (help, legal, .). We will use
MySQL as our database, which has some search functionality integrated
but
only with MyISAM tables and not InnoDB. It also fails with phonetic
searches. In Java I would use Lucene fot that which is a great search
engine. Does anybody know about equivalent search technologies or does
anybody has solved equal problems?
The same is with CMS - in the Javaworld you have Alfresco, OpenCms, . -
are
there any good Ruby/Rails implementations out there?
Thank you for your help.
Mirko
Mirko N. wrote:
One of the key
advantages is that new developers will understand this platform much
quicker
than all the Java EE Apis and Frameworks.
Amen!
n Java I would use Lucene fot that which is a great search
engine. Does anybody know about equivalent search technologies or does
anybody has solved equal problems?
Ferret is a ruby port of lucene (the indexes are compatible)…
http://ferret.davebalmain.com/trac/
Mikkel
On Jan 23, 2006, at 3:26 AM, mikkel wrote:
n Java I would use Lucene fot that which is a great search
engine. Does anybody know about equivalent search technologies or
does
anybody has solved equal problems?
Ferret is a ruby port of lucene (the indexes are compatible)…
Unfortunately that is not entirely true. I’ve tried using indexes
built with Java with Ferret and had show stopping issues. If the
indexes are built entirely with Ferret then they’ll work with Ferret,
but interoperability (which is very important for my application)
currently is not possible (in my experiments).
Erik
On Jan 24, 2006, at 3:04 PM, Mirko N. wrote:
What are your experiences with Ferret in comparison to Lucene
(which I like
a lot)?
My Ferret experiences are still light, primarily because of the
incompatibility with Java Lucene which is a show stopper for me to
proceed with my application using Ferret. Ferret is nicely done,
don’t get me wrong. Kudos to Dave in a major way!
But it is not scalable enough yet, and I’d say not ready for prime-
time for systems with a substantial amount of data. Dave’s cFerret
promises to fix all of the performance issues. And I surely hope
he’ll consider making indexes compatible with Java Lucene (or helping
Java Lucene change its ways, as the case may be).
documentation
looks good.
I’m using Java Lucene via an XML-RPC custom server and its working
very nicely. Super fast thanks to the incredible speed of Java
Lucene. As long as you are careful about what goes over the wire
with remote calls to Lucene, it can be a great solution.
I certainly am heavily biased towards Lucene, of course. But other
solutions like TSearch2 are very likely not as flexible as doing
custom work with a Lucene index, and more than likely not as fast
either.
Erik
On 1/24/06, Mirko N. [email protected] wrote:
I’ve no experience about the performance of TSearch2, but the documentation
looks good.
I tested TSearch2 in a search engine for about 450 large text files
(~50MB total) for a site. The documentation is good and everything
works well, but it was quite slow. This was using an untuned
PostgreSQL 7.4 installation, so it is possible that it works much
better with 8.1 and some tuning.
I also tested Swish-E and ht://Dig. Swish-E was good but doesn’t do
excerpts, so I really couldn’t use it. I ended up going with
ht://Dig, which performs well and does excerpts. ht://Dig was a
couple orders of magnitude faster than TSearch2 on the same hardware
using my dataset.
I can echo the kudos to Ferret. It is a project with great potential.
I’m
developing with it right now, and haven’t gone to production, so I can’t
speak to scalablity, but its api is close enough to Lucene’s that it was
easy for me to switch over (actually, I come from pyLucene, which is
another
amazing project).
I believe that Dave’s goal is to follow the Lucene api, so I’m ok with
using
Ferret while its still in its early stages, because the api is mostly
stabilized.
matt
On 1/24/06, Mirko N. [email protected] wrote:
- Using PosgreSQL with TSearch2
Thanks.
Unfortunately that is not entirely true. I’ve tried using indexes
http://lists.rubyonrails.org/mailman/listinfo/rails
Rails mailing list
[email protected]
http://lists.rubyonrails.org/mailman/listinfo/rails
Any feedback, feature requests or patches for RailsCron are more than
welcome! I’ll look into Quartz, and see what they’re offering too.
–
Kyle M.
Chief Technologist
E Factor Media // FN Interactive
[email protected]
1-866-263-3261
Hi Erik,
What are your experiences with Ferret in comparison to Lucene (which I
like
a lot)?
I am currently thinking about these alternatives to Ferret/SimpleSearch
(which are the two possible choices for Ruby I’ve found, both seem to be
not
as mature or feature rich as Lucene):
- Calling Lucene as a Webservice from Ruby
- Using PosgreSQL with TSearch2
I’ve no experience about the performance of TSearch2, but the
documentation
looks good.
I am also looking to an alternative to Java’s Quartz project - I’ve
found
RailsCron, but this is still a 0.1 and with not so much functionality.
Does
anybody have some alternatives to that?
Still searching for a RoR CMS…?
Thanks.
Mirko
-----Ursprüngliche Nachricht-----
Von: [email protected]
[mailto:[email protected]] Im Auftrag von Erik H.
Gesendet: Dienstag, 24. Januar 2006 20:51
An: [email protected]
Betreff: Re: [Rails] Re: Search functionality and CMS
On Jan 23, 2006, at 3:26 AM, mikkel wrote:
n Java I would use Lucene fot that which is a great search
engine. Does anybody know about equivalent search technologies or
does
anybody has solved equal problems?
Ferret is a ruby port of lucene (the indexes are compatible)…
Unfortunately that is not entirely true. I’ve tried using indexes
built with Java with Ferret and had show stopping issues. If the
indexes are built entirely with Ferret then they’ll work with Ferret,
but interoperability (which is very important for my application)
currently is not possible (in my experiments).
Erik
Rails mailing list
[email protected]
http://lists.rubyonrails.org/mailman/listinfo/rails