Etherdev Tumblr

Having spent hours digging into this issue with little help from the getshopped forums, I figured I should share this. I have a client who is using Nginx as web server and we're hosting a Wordpress site along with some Rails apps from the same server. Now, Wordpress itself plays perfectly fine with Nginx assuming you get the configuration right (see posts from Chris Kempson and the Nginx wiki for that). The biggest issue is really centered around permalinks, which a simple little plugin called Nginx Compatibility typically deals with quite nicely.

Unfortunately, WPEC doesn't play well with this setup (at least as of this writing using WP e-Commerce 3.8.7.2). If you try to use "pretty urls" of any kind, WPEC will attempt to append index.php to any of its AJAX calls (use Firebug console to see the POST) which results in a 404 error on those calls.

The only solution (or, a hacky compromise that I HATE, really) that I've been able to come up with so far is to disable Nginx Compatibility and set permalinks to something like the following:

/index.php/%category%/%postname%/

(Update the permalinks settings TWICE, and then flush the WPEC Theme cache)

You should be able to use any variety of permalink there, but the critical part is the /index.php/ slug. It's ugly. And as I said, I hate it. But it does keep WPEC from applying index.php to calls and suddenly you can update your cart, clear the cart, etc.

Hybridizing My App Design Approach

Interesting to see how Ryan Singer of 37 signals approaches app design from the perspective of UI. Granted, he's mostly a UI designer, not a back-end programmer (in fact, he specifies where in the process he hands it off)... but, this is still oddly similar to how I worked several years ago before we started trying TDD/BDD. Of course, I was also more the UI guy between the two of us.

Recently, I've put UI off until after writing a lot of tests on the model. And after that is exactly were I feel stuck.... lost on trying to envision Cucumber features before being able to see anything tangible. My brain _does not like_. Now, I think I'm going to try and merge the two approaches... rough-sketch basic model relationships, wireframe really raw UI, translate that into simple HTML templates, THEN start spec'ing the models... then describe the sketches in Cucumber features and finally start implementing code. That makes way more sense to me. I feel like I really need those steps of sketching wire-frames and getting a static version in the browser before I can fully envision the integration tests.

This is a great article by John Nunemaker about the benefits of using MongoDB in Ruby on Rails. I've wanted to mess around with document-oriented databases for a while (we were considering it with Capocus! 2.5 back in early 2009 even, but it felt too new). And since I'm starting from scratch with Capocus! 3 in Rails 3.1, I figure there's no time like now. I've been going back and forth on CouchDB vs. MongoDB. Since I'm hoping to do a good bit of iOS app tinkering for apps that work in conjunction with some Capocus implementations, CouchDB has some appealing features (REST, data sync across devices, small size for running on smartphones). But this article may have tipped me toward starting with MongoDB. It seems to have a more vibrant user base in the Rails community, and John's examples here are pretty exciting. And since Capocus! will likely be using a lot of internal stats logging, the need for better write performance might win out over MVCC). I may spend a day or two with each and see how it shakes out.

excerpt from Drive, by Daniel H. Pink

*Harry F. Harlow was a professor of psychology at the University of Wisconsin who, in the 1940s, established one of the worlds first laboratories for studying primate behavior.