My first formal introduction to TDD was a three-day TDD workshop put on by Jim Weirich and Joe O’Brien while I was at the Los Angeles Times. Jim, an expert EMACS user, had his system setup to run tests quickly from within the editor, even showing a red or green notice indicating the pass/fail status of the test.

This may not seem like a big deal to people using any number of IDEs but personally, I’m a Vim user (specifically MacVim, it’s an extremely well done port of Vim using Cocoa). I love it. I’m way faster editing inside of Vim; my brain has mapped around all of the commands because I’ve been editing with it for so long. I’ve been using vi variants in one form or another since my Freshman year of college when my Computer Science 101 professor forced us all to code with it for our programming projects on the school’s Sun server. While doing Java work two or so years ago, I developed mostly on Eclipse and eventually broke down and purchased a license for the Eclipse vi keybinding plugin. Now that I’m employed for Ruby work I’ve switched back to Vim full-time (after a 1/2 hour dabbling with Textmate).

Jim’s EMACS setup made me jealous. During the class I figured out how to set the make program to use ruby (or spec for rspec tests) and Quickfix will run the tests for the file. Here’s the command:

:set makeprg=ruby\ %

The \ escapes the space and the % gets expanded to the current buffer’s filename when Quickfix is activated. Quickfix was originally used for a “quick” way of compiling your program to see if there’s any errors. Vim will even move your cursor to the file and line number of the first error if there are any. This works for ruby tests too! Another Vim setting, errorformat, is how Vim parses errors from running makeprg.

OK, so this is all great, but what if my test file is huge? What if it takes 10+ seconds to run? What if I’m only interested in a single test because I’m being a good developer and practicing TDD?

This is where the Ruby Single Test plugin comes in. The plugin is an extraction of some functions I’ve been playing with in my .vimrc file for the past couple of weeks. All it does is provide a command for running the test block your cursor is hovering over in your test file (<leader>. by default.  <leader> is usually comma or backslash). Most of the time you’re only interested in a single test while you’re editing code so turn-around time is vastly improved. No more jumping back to the console to run a test.

To install just drop the ruby_single_test.vim file in your .vim/plugin directory.

Currently it works for Test::Unit as well as Rspec tests.  I’ve been using this quite a bit recently and I think I’ve worked out most of the bugs. My setup isn’t necessarily the same any everyone’s so please let me know if you have any problems with it! The code is up on Github, so feel free to play with it.

Inplace_preview

Inplace_preview is the first. It monkeypatches Scriptaculous‘ InPlaceEditor to give it a preview mode. It also includes an ability to do an ajax call for rendering the preview. It comes in handy when using an InPlaceEditor for editing HTML but with the ajax rendering anything would be possible.

Here’s an example (if you save you’ll get a big-ol error since we don’t actually save here):

Click Here to Edit…

Gallery_scroller

Gallery_scroller is small Javascript library for creating a scrolling gallery that we’re using on more projects then I expected. It requires Prototype and Scriptaculous effects.js libraries.

Here’s an example:

 

Barack Obama says John McCain will…

 

Barack Obama praises Colin Powell

 

Barack Obama accepts the…

 

Barack Obama’s faith-based…

 

Barack Obama campaigns in Wisconsin

 

What went right for Barack Obama

 

Barack Obama criticized for…

 

Barack Obama officially becomes…

 

Barack Obama joins Joe Biden at…

 

Democrats nominate Barack Obama

 

Barack Obama addresses the NAACP

 

Barack Obama on personal…

 

Preparations underway for Barack…

 

Barack Obama promises tax cut if…

 

Barack Obama accepts Democratic…

 

Barack Obama condemns rumors…

 

Barack Obama and John McCain urge…

 

Barack Obama’s relationship with…

 

Barack Obama

 

Before Barack Obama: Alternative…

 

Barack Obama’s funding advantage…

 

Barack Obama’s seal controversy

 

Barack Obama and Hillary Clinton…

 

Barack Obama and Muslim voters

 

Barack Obama on Hillary Clinton

 

Barack Obama and Scarlett Johansson

 

Pop culture candidates

 

Democratic debate

 

Super Tuesday: Nation

The documentation on both of these goes into detail on how they work, so check out their github sites!

Reid also posted on Gem Install That about another plugin we worked on, Subselector, which extends ActiveRecord to give it subselection capibilities.

In a nutshell cached_resource automagically caches ActiveResource response objects in a memcached instance.  The plugin requires cache_fu for memcached configuration goodness.  See also: Kickin’ ass with cache-fu.

The code for cached_resource is also up on GitHub along with a many other Los Angeles Times plugins and other tidbits.  You can find it here: http://github.com/latimes/cached_resource/tree/master

So, how do you get it to work?  After dropping the plugins in vendor/plugins, set up your cache_fu configuration then drop this single line in any of your ActiveResource models:

class MyClass < ActiveResource::Base
  cached_resource
end

The cached_resource call also takes normal cache_fu options:

class MyClass < ActiveResource::Base
  cached_resource :ttl => 15.minutes
end

So now every time you run a .find() on MyClass the results will be cached for the next time you do the same kind of find.  The cache key is the URI of the ActiveRecord generated REST query.

In addition, if you do MyClass.find(:all) or the like to get a list of resources, individual instances in the returned list will be cached separately and the list itself will be distilled down to IDs before it’s stuffed in its own cache bucket.  When pulled from the cache the list will be reconstituted by pulling the individual objects from the cache by ID or making additional ActiveResource queries to get objects that have fallen out for one reason or another. The upshot of this is you’ll have less of a chance of stale data between mixed finds.

The cached_resource plugin makes some assumptions about how your REST service is set up.  If the REST service you’re pointing to with ActiveResource are wacky, i.e. not conforming to exactly what ActiveResource expects its URLs to be, this plugin will probably not work for you.  That said, if you can do normal finds with IDs, like MyClass.find(123), you’ll probably be fine.

This plugin was written before memcached support was baked into Rails 2.1 so if anyone wants to fork and get it working please be my guest!  I’ll be happy to pull it back in

Credit where credit is due: Josh Kleinpeter originally came up with the cached_resource concept and Subba Rao helped to lay the groundwork.

This one is called Craken.  In a nutshell it manages and installs rake-centric cron jobs.  Coupled with Capistrano goodness it is the answer to your recurring task needs in Rails.

It’s up on GitHub: http://github.com/latimes/craken/tree/master

Here’s the deal: with many web applications comes the need to do recurring tasks on the servers the application runs on.  These sorts of tasks can range from running backups, clearing caches, emailing notices and other such maintenance-y things.

Who manages these tasks?  Ideally the task code as well as configuration and timing should be kept with the source code of the application itself and not as part of some further set of scripts that reside in a dubious location in your source control.  This is what Craken does.

The tasks are defined as rake tasks within the application.  Craken itself is a rake task that sets up a crontab for defined tasks based on your current application configuration.  The plugin also includes a Capistrano recipe that runs the Craken rake task on remote servers making deployment a breeze.

Craken, by default, looks for recurring task definitions here:

RAILS_ROOT/config/craken/crontab

You can also define tasks for a particular machine by prefixing the machine name (separated with an underscore):

RAILS_ROOT/config/craken/MyMachineName_crontab

The file format is modeled after crontab:

30 1 * * * thing:to_do

In this case thing:to_do is a rake task.  It will run at 1:30 AM every morning (take a look at the crontab format if this makes no sense).

The command to install the cron job locally:

% rake craken:install

If you’re using Capistrano for deployment (and you really should) a recipe is provided to run craken:install on remote machines.  Your Capistrano configuration needs to define a :cron role so the recipe knows which machines it should run the command on.

% cap craken:install

Ideally you can make craken:install part of your deployment configuration, :after_deploy would be a good place to put it.  Craken overwrites the existing configuration for the application on the machine so running it multiple times is not a problem.  Running multiple apps on the same machine or cron jobs placed by hand?  No worries, Craken delimits the lines it generates in crontab with comments and doesn’t disturb existing entries.

For more information about the plugin, peruse the README file.

If you have any questions/comments/suggestions about Craken please leave a comment!

The name “Craken” generously suggested by Giles Bowkett.

The LED Flowerpot!

This is a little project I built as a gift for the Wife for the combined 5-year being together anniversary/Valentine’s Day. I did most of the work on the weekends, soldering and gluing in the back yard out of sight from prying wifey eyes.

The circuit is not terribly complicated, it’s a modified version of Wilf Rigter’s Quad LED Fader circuit (specifically QLF6) I found on the BEAM robotics community server. It uses a 74C14 Schmitt triggered inverter chip as a bank of five oscillators fed into a LM339 quad comparator to generate four slow wave forms to drive four banks of LEDs. The 74C14 was provided by my good friend and chip dealer, Mr. Bunt. The major addition I made to the circuit was a voltage regulator to change its power supply to use a wall wart scavenged from a defunct answering machine that I haven’t seen in years. A 9-volt battery, its package being the ideal size to fit in the flowerpot base, unfortunately drained quickly with the number of LEDs I wanted to use.

Soldering the circuit itself onto the prototype board wasn’t much of a chore but as you can see, I do needs me some practice at circuit layout. The most irritating thing was wiring up all the LEDs. There’s 34 of the little buggers and each one had to be hot glued in and wired up to their brethren and the circuit. The result of which is the rat’s nest you see before you. I only have black wire, OK?

The prototype board rests on the sides of the pot, stuffed down below the “dirt’s” surface. The surface itself is made from a round cut from a cheap cutting board that was too crappy to actually use as a cutting board. Scraps left over from the same were hot glued to the sides of the pot to hold the surface at the desired depth.

The LEDs were hotglued into position before being wired up for action. First I had to make holes in the cutting board round to stuff the LEDs through. I originally intended to use a drill to create the holes, but quickly found that my smallest drill bit opened cavernous orifices compared to the width of the leads on your average LED. In the end I used a needle heated with a lighter to pierce the holes. Not only was the pierced diameter perfectly suited for LED leads, I found the process of holing plastic with a glowing-hot needle extremely satisfying for some strange reason. It did take a while though.

After the LEDs were in place and the circuit was working, the next step was to decorate the “dirt”. I used soil-colored foam turf I found at a model railroad store along with “scenic cement” to fix it in place. The added details were bits of bark and very small rocks I found in our real plant pots and about the yard; without them the scene didn’t look terribly realistic. Not that LEDs sprouting out of a fine soil-filled pot is firmly grounded in reality or anything.

When I turned from using a 9V battery fueling the circuit to a wall wart, I had intended to drill a hole in the side of the pot near the bottom to pass the power cable through. After a couple of attempts, the near hideous scratching of the pot’s green glaze and a paralyzing fear of causing it to shatter I decided to use the pot’s drain hole as mother natured intended. Its hole-ness beckoned to me. Besides the socket I had fit almost perfectly. Unfortunately the pot can’t rest on its base anymore; it requires I use a stand as you see in the first picture. The current stand is a opened leatherman knock-off because it was the first thing I had on hand that fit the bill. It was a poor knock-off so it makes a better stand anyway.

Since pictures don’t show off its glow-y goodness, here’s a video I took of it doing its thing (in the dark so you can see it better):

So where did I get the idea to do this? I was playing around with the circuit on a breadboard and a little garden of LEDs glowing and fading in and out when the wife came by and said “Ooooo! Mushrooms!”. The idea for the finished product was a short hop away from there.

The crust for this pie was quite different then what I’m used to: it was a shortbread crust complete with heart-stopping amounts of butter. Crumbly, but oh so tasty. I’ve had the worst luck making pie crusts but this one worked out fairly well apart from making a terrible mess. Tonight I took the rest of the dough, rolled it into a log, sliced it into half-inch pieces, sprinkled it with lavender sugar that the wife had picked up, and baked it in a 350 degree oven for around 18 minutes. Leftover pie crust shortbread cookies!

I’ve been enjoying the new Jamie at Home show, it has a quiet calmness about it due to the refreshing lack of background music. The one insanely irritating thing is that they don’t publish all of the recipes on the web site. The pie episode we were watching had a fabulous recipe for Steak, Guinness and Cheese Pie but the recipe is nowhere to be found. The deplorable state of the Food Network’s website in general is a rant for another day. The only mention of the missing recipes (besides the ranting in the available recipe comments) was this blurb on the show’s main page:

Please note that only two recipes from each Jamie at Home episode will appear online. Also, access to recipes is limited to users within the United States.

A little explanation would be nice!

In any case the web comes to the rescue with the missing recipe. Yay interwebs!