Embedded full-text search

AJ ONeal coolaj86 at gmail.com
Wed Jun 6 12:21:50 MDT 2012


I realize that neither SQL nor NoSQL have the solution here - unless I use
something that would be easy to store each word / prefix to index on. There
aren't 1,000,000 words in the set of strings, probably just 5,000 plus
another 1,000 misspellings.

If something like Lucene, Solr, Sphinx were written in Go or Lua that would
probably be the best choice. Sphinx is probably the most light-weight of
the two (Java platforms tend to always have 10,000 lbs of gorilla nonsense
included)

Storing everything into a file and running grep will probably be as fast as
or faster than SQL.

AJ ONeal

On Wed, Jun 6, 2012 at 12:01 PM, Steve Meyers <steve-plug at spwiz.com> wrote:

> On 6/6/12 11:42 AM, Eric Jacobs wrote:
> > Definitely make sure that you're using an index on your table. That
> > will probably speed up your queries quite a bit. I have no idea what
> > you would index on this table though.
>
> Normal indexes do nothing for full-text searches.  You need a full-text
> index of some sort.  MySQL has support for those, but it's not really
> all that great.  There are much better full-text search engines out
> there, like Lucene or Sphinx.  I don't know how well either would do on
> a Raspberry Pi.
>
> Steve
>
> /*
> PLUG: http://plug.org, #utah on irc.freenode.net
> Unsubscribe: http://plug.org/mailman/options/plug
> Don't fear the penguin.
> */
>


More information about the PLUG mailing list