Hi, Gernot! Thanks for your answer.
> Please note that running a production configuration with the direct
> writer is dangerous unless you are absolutely sure that you have only
> one instance of your app running (see the README for details). I would
> recommend to always use the beanstalk writer in production
I know! At the moment beanstalk isn't installed in my hosting, and the
site is in a pre-production environment, so in the meantime I'm using
the direct writer.
> Since you know xapit: Do you miss any configuration options for the
> xapian engine?
I haven't used the configuration options heavily, but I'll make some
general comments.
Using xapit, I could write something like:
@posts = Post.search "xapian", :per_page => 15, :page => 1
@posts.each { |post| link_to post.title, post }
Same code with xapian_db:
@results = Post.search "xapian", :per_page => 15, :page => 1
@results.paginate.each { |result| link_to result.title,
result.indexed_object }
I find xapit's way more elegant to do such a common task.
In the README you recommend configuring the blueprints in a separate
config file. I'm not doing it that way. The reason is, I'm using Spork
as a test server. If I include the blueprints for my Post model in the
config file, then that model is cached when Spork is started. The
simplest solution I found was moving the blueprints to the Post model.
Right now I don't remember if I used xapit or acts_as_xapian, but
there was an application where I used a xapian gem to find similar or
related posts to the one I was showing.
And finally, I prefer xapit's way of returning a non-existent
spelling_suggestion as nil instead of as an empty string, so it's
easier to use as a condition. But that's a matter of taste, I guess.
I've just started using it, so if in the future I've got more
comments, I'll let you know.
Regards.