[ANN] Sphincter 1.0.0 Released


#1

Sphincter version 1.0.0 has been released!

http://seattlerb.org/Sphincter

Sphincter uses Dmytro Shteflyuk’s sphinx Ruby API and automatic
configuration to make totally rad ActiveRecord searching. Well, you
still have to tell Sphincter what models you want to search. It
doesn’t read your mind.

Features:

  • Automatically configures itself.
  • Handy set of rake tasks for easy, automatic management.
  • Automatically adds has_many metadata for searching across the
    association.
  • Stub for testing without connecting to searchd, Sphincter::SearchStub.
  • Easy pagination support.
  • Filtering by index metadata and ranges, including dates.

Problems:

  • Setting match mode not supported.
  • Setting sort mode not supported.
  • Setting per-field weights not supported.
  • Setting id range not supported.
  • Setting group-by not supported.

See README.txt for quick-start, examples, etc.

Changes:

1.0.0 / 2007-07-26

  • 1 major enhancement
    • Birthday!

http://seattlerb.org/Sphincter


Poor workers blame their tools. Good workers build better tools. The
best workers get their tools to do the work for them. – Syndicate Wars


#2

Hey Eric,

Looks pretty awesome. Can you tell me the main differences beteween
this and say the acts_as_sphinx plugin? We’ll be looking into
implementing site search in the coming months and I’m fairly sure I
want to use Sphinx (awesome stuff).


#3

On Jul 30, 2007, at 22:34, Marston A. wrote:

Hey Eric,

Looks pretty awesome. Can you tell me the main differences beteween
this and say the acts_as_sphinx plugin? We’ll be looking into
implementing site search in the coming months and I’m fairly sure I
want to use Sphinx (awesome stuff).

I set out to write something that would take away as much of my work
as possible, and I left out anything I didn’t need. Some people
might want those things and I designed it to be easy for most people
to add them (and tested it to make sure). I picked what I thought
were sensible defaults for configuration options and automated it as
much as possible.

acts_as_sphinx does much less than Sphincter, which means you have to
do more.

With acts_as_sphinx:

  • You have to write the configuration file by hand.
  • You have to build the source queries by hand.
  • You don’t get easy searches via associations like
    my_blog.posts.search for free.
  • You’re hard-coded to /var/run, which probably won’t work for VPS/
    slice setups.
  • You need a separate running sphinx daemon to test searching.
  • You don’t have any test coverage (but acts_as_sphinx is tiny, with
    164 lines of code.)

With acts_as_sphinx, you probably get some of these that Sphincter
doesn’t support:

On Jul 31, 12:20 am, Eric H. removed_email_address@domain.invalid wrote:

Sphincter version 1.0.0 has been released!

Problems:

  • Setting match mode not supported.
  • Setting sort mode not supported.
  • Setting per-field weights not supported.
  • Setting id range not supported.
  • Setting group-by not supported.

But I bet you don’t need those anyhow, and if you did, they’ll be
easy to add.

I started with ultrasphinx, but I couldn’t understand how to set it
up to do what I wanted, I couldn’t read the code, and it would’ve
taken me longer to untangle it enough to make it do what I wanted
than to build something from scratch.

With ultrasphinx:

  • You don’t have full automatic configuration. There’s some
    automatic configuration, but its in the sphinx.conf format rather
    than handy yaml, and its less automated.
  • You don’t get tested code. (ultrasphinx is about 2x the LOC as
    Sphincter and has about 3x the magic.)
  • You don’t get easy searches via associations like
    my_blog.posts.search for free.
  • You don’t get daemon-free testing of searching.
  • You do get spelling support (but I can’t see where it is tied in)
  • You do get excerpting
  • You do get scary warnings from searchd about missing indexes.


Poor workers blame their tools. Good workers build better tools. The
best workers get their tools to do the work for them. – Syndicate Wars


#4

Do you have any comparison between this and acts_as_solr? That’s
pretty simple to use as well.

Thanks


#5

On Aug 1, 2007, at 10:02, s.ross wrote:

Do you have any comparison between this and acts_as_solr? That’s
pretty simple to use as well.

No, solr looks a lot bigger than Sphincter. It also requires the JRE
which is a lot more files than just Sphinx.

Looking at its acts_as_solr method, Lucene has many more features in
its search engine. You also get on-the-fly indexing, rather than
having to use a cron job to update sphinx’s index.

Rumor has it that reindexing using Sphinx takes less time than with
Lucene, but I wouldn’t know because I’ve never compared the too. It
could be all lies.


Poor workers blame their tools. Good workers build better tools. The
best workers get their tools to do the work for them. – Syndicate Wars


#6

On Aug 1, 2007, at 11:32 AM, Eric H. wrote:

its search engine. You also get on-the-fly indexing, rather than
having to use a cron job to update sphinx’s index.

Rumor has it that reindexing using Sphinx takes less time than with
Lucene, but I wouldn’t know because I’ve never compared the too. It
could be all lies.

I’ve seen sphinx index the same dataset in under 2 minutes that it
took solr 20 minutes to index. Solr does have more features, but boy
sphinx is fast and easy to work with.

Cheers-
– Ezra Z.
– Founder & Ruby Hacker
– removed_email_address@domain.invalid
– Engine Y., Serious Rails Hosting
– (866) 518-YARD (9273)