An extremely informative and well-designed site. I wish all open source projects had a section which explained why they chose the components that they have chosen and a high level diagram of how the technology works.

I agree, this is really well designed. But it seems to lack an easy to access link back to the github repo. It has a giant download link, but who wants to download and manage versions when they can clone a repo?

You are right, we corrected this.

Thanks!

Why do you recommend easy_install (vs pip)? I have hardly any experience with easy_install so I'm always curious when projects use it.

http://stackoverflow.com/questions/3220404/why-use-pip-over-...

Project looks great. Great Github page too – is this a stock Github template?

Thanks! It's a custom design. The designer's page is http://danaepanopoulou.com/

It's really beautiful, thanks for the link!

This could be very powerful if it gets:

Clustering & distributed indexing/searching

sphinx performance per node is amazing but it lacks clustering and if that comes along it will put serious pressure on ElasticSearch.

For faceting, there's the fSphinx library:

https://github.com/alexksikes/fSphinx

Thanks, I'll look further into that project.

Or, you know, just use elasticsearch.

Sure, elasticsearch is absolutely great, no question on that. Nevertheless, a) it requires running Java and configuring JVM, b) Sphinx full-text search engine is really powerful, reliable and fast. So why not build a handy interface around Sphinx also? There is always room for more open-source software :).

Wouldn't this have a lot of moving parts as compared to Elasticsearch? Redis, django, nginx, sphinx all have to be configured. Elasticsearch already has all the feature you list as coming soon, and I imagine it would be very hard to add transparent clustering to this setup.

Seemingly yes. My most basic argument on that is that these components are commonly found in modern server stacks, so chances are that you are already using them, you know how to configure and setup. If I am already using Nginx & Redis and Sphinx as a full-text search engine, why not have something to facilitate the search in modern manner, almost out-of-the-box? As far as the clustering part you are referring, I don't have an exact scenario on my mind at the moment but I believe it can be done. Please keep in mind that this is the very first beta version and hopefully with several contributions, it will undergo many significant changes.

I notice that MySQL is being used, is that a hard dependency or can/will we be able to use Postgres?

In order to handle realtime indexes Sphinx requires using SphinxQL which in turn uses MySQL client protocol. So you'll need the MySQL client package. MySQL server is not strictly required and I am guessing can be replaced by Postgres since Django models make the DB backend transparent to the developer.