NYCPHP Meetup

[Eddie Drapkin]

On Thu, Jan 21, 2010 at 1:09 PM, Jake McGraw <jmcgraw1 at gmail.com> wrote:
> On Thu, Jan 21, 2010 at 1:05 PM, Matt Juszczak <matt at atopia.net> wrote:
>>> InnoDB for everything but tables that require fulltext search, which
>>> is not yet supported on InnoDB.
>>
>> Can always use something like lucene for this.
>
> Yea, but then you've got to keep a Tomcat instance up and know Java or
> use a PHP implementation which can be slow. MySQL Fulltext works
> really well in most instances.
>

That's just a little bit inaccurate.  While it certainly helps to know
Java, running a Solr instance requires only the ability to start a
java process and read some documentation.  The entire thing is
communicated to / from a REST API and it's configured with (over|very
well) documented XML files.  There's no knowledge of Java or lucene or
tokenizing or anything other than how to read documentation required.
It will almost always work out better than MySQL FULLTEXT because Solr
has some niceties built into it, like search suggestions, spelling
corrections, stemming, etc.  There's a lot of other cool stuff built
into it now, like document importing (I think they support PDF, MS
Word, OO.o formats, etc. but I'm not too sure as I don't use that
feature myself), multi-core searching, there's a bundled tool for
indexing out of SQL and a lot more I'm sure that I'm forgetting. You
can go from "no search" to "well implemented, professional feeling
search" in two or three days, with no prior experience in Java or
advanced text searching.