Binary Code

I use GeekTool to display quite a few items on my desktop, it’s a fantastic utility. For fun I’ve put together a Ruby script for use with GeekTool that displays iCal events: GeekCal.

Download the script, make it executable, set it up as a shell script in GeekTool et voila. Calendar events on the desktop.

Seems its possible to crash the current version of rcov hard under certain circumstances (version as of this writing: rcov (0.8.1.2.0)). If you find it crashing with:

ruby(17057,0xa0124720) malloc: *** error for object 0×5b1a602: Non-aligned pointer being freed
*** set a breakpoint in malloc_error_break to debug
/opt/local/lib/ruby/1.8/rss/rss.rb:274: [BUG] Segmentation fault
ruby 1.8.7 (2008-08-11 patchlevel 72) [i686-darwin9]
…
rake aborted!
NaN

Check to see if you have any empty helper files that don’t have any coverage. I had a couple that resembled the following. Removing them removed the “divide by zero” issue that causes the crash:

module ProductHelper
end

Update
Scratch that, it doesn’t entirely fix the issue. I’ve also implemented Stew Welbourne’s hack for round_to_tenths(decimal) in MetricFu but that too does not seem to be a complete solution. The seg fault still happens, though intermittently (sometimes it runs through the complete test suite, sometimes it doesn’t).

The search continues.

Introducing mysql2sqlite.rb, a Ruby script for converting MySQL databases into Sqlite databases.

Based on my googling of the web, I figure I’m one of perhaps five people in the world who’ve ever wanted to do this conversion but in case you’re number six, here you go.

Usage is pretty straight-forward. It can either be configured via command-line parameters or via a specified YAML file:

./mysql2sqlite.rb database_name database_user database_password
./mysql2sqlite.rb config.yaml

and it will spit out two files: one of the raw, converted SQL and one that is the Sqlite database. For instance, if your MySQL database was called ‘clients’ you’d end up with clients.sql and clients.sqlite.

For more details and source code see the mysql2sqlite.rb project on github.

Last month I started rewriting the admin reporting section of one of my sites in Merb rather than refactor the existing PHP version, for a number of reasons. The pace of development with Ruby and Merb was tremendous and within about a week and a half I had a fully-functional, extensible reporting site developed, replete with graphs and statistics. DataMapper made tying into the legacy database, with it’s rather novel table and column naming schemes, trivial. It was all tremendously satisfying and, dare I say it for a project so unglamorous, fun.

And then I tried to get it to run on our production server, which is running Red Hat Enterprise Linux 5, and the fun went away.

I first tried installing Phusion Passenger (a product I’m absolutely enthralled by) but no such luck; it had a hell of a time with the default Apache install and lack of development headers and various file locations. I wasn’t willing to muck about with the production server in a vain attempt to force it to work at the risk of the rest of the site.

Instead I figured it ought to run just fine on Rack. And every dependent gem installed with nary a hitch save for DataMapper, specifically do_mysql, which would fail to build the native extension with the following error:

In file included from /usr/include/mysql/my_global.h:83, from do_mysql_ext.c:6:
/usr/include/mysql/my_config.h:15:28: error: my_config_i386.h: No such file or directory

The solution was ultimately provided by Dan Kubb:

We probably should make it so that all the DO driver specs can be run on installed gems, but in the meantime, checkout the source from git using the following commands:

git clone git://github.com/datamapper/do.git
cd do/data_objects
sudo rake install
cd ../do_mysql
... remove the references to my_config.h in do_mysql ...
rake compile spec
sudo rake install

This will first install the edge version of DataObjects, and then will compile and run the specs for do_mysql, and then install it. You will want to remove all references to my_config.h from the do_mysql C libs just prior to running the specs of course.

The really important bit here: “remove the references to my_config.h in do_mysql”. That’s the magic, and with that everything was good and right and Merb was run, and the users were happy.

(For the very curious, the entire DataMapper Google Groups thread.)

As part of the deployment of my first Merb app I wanted to make sure it contained everything it would need. This is pretty easy in Merb:

thor merb:gem:install

Once that runs through and bundles all the gems in your dependencies.rb file I knew that I should have been able to simply run:

./bin/merb

and use the bundled install of Merb. But this was failing mightily with dramatic FATAL errors. Using:

./bin/merb --verbose

shed light onto the issue. It turns out that because I was doing all my development by just running merb Merb was loading necessary dependencies that I hadn’t specified explicitly in my dependencies.rb file. Dependencies like “rmagick” and “mongrel”.

Putting all the dependencies into the file and re-running thor merb:gem:install did the trick.

Incidentally, if you attempt to do something like:

thor merb:gem:install mongrel

without having mongrel specified in dependencies.rb the install will fail. This, in hindsight, makes perfect sense to me – I like that file being the authority – however the error message generated is a bit obtuse ( in this case: "Configuration could not be confirmed: Could not find RubyGem mongrel (>= 0)").

It also appears (though I’m surprised by this so I suspect my understanding of this process is suspect) that you’ll need to declare dependencies that dependent gems rely on. Again in my case to get gruff bundled I also had to define rmagick:

dependency "rmagick", "2.9.0"
dependency "gruff", "0.3.4"

So if you’re bundling gems make sure they’re all explicitly defined and use –verbose when running Merb to double-check. And read Getting Started with Merb – Bundling Merb with your Application. It was immensely helpful.

I believe this should be taken as dogma and followed religiously:

I reject the argument because I reject as buggy any code such that object o responds to method m but o.responds_to?(:m) => false. If you implement your own method_missing for a class, you should almost always implement your own responds_to? as well.

- Maybe

There is wisdom in them there words. Monkey-patching is (perhaps) ok, but voodoo should probably be avoided at all costs and an object that can act but won’t provide a means of letting you know without actually trying it is just plain ol’ voodoo. And programming voodoo is bad juju.

You’re installing Ultraspinx (perhaps per these instructions from Inoshi, which are the best I’ve found thus far) and you run into this error when time comes to bootstrap your installation:

~/Sites/ticklists/: sudo rake ultrasphinx:bootstrap –trace
(in /Users/chris/Sites/ticklists)
** Invoke ultrasphinx:bootstrap (first_time)
** Invoke ultrasphinx:_environment (first_time)
** Invoke environment (first_time)
** Execute environment
** Execute ultrasphinx:_environment
** Invoke ultrasphinx:configure (first_time)
** Invoke ultrasphinx:_environment
** Execute ultrasphinx:configure
rake aborted!
You have a nil object when you didn’t expect it!
You might have expected an instance of ActiveRecord::Base.
The error occurred while evaluating nil.[]

I couldn’t find this one in the google anywhere. Turns out the cause of this is defining non-existent properties for indexing on your models (perhaps I’m the only one dumb enough to do that).

You’ll know if this is the cause because you’ll see something akin to this in your Rails console:

** ultrasphinx: warning: field f is not present in User
** ultrasphinx: warning: field descritpion is not present in Profile

Remove (or fix the spelling of) those and everything should work hunky-dory.

A couple of weeks ago, in the middle of getting DreamBank ready for launch I was feeling like I needed a break. Of course that meant “what can I develop in Rails over a single weekend?” See, I had an idea for an application that would fulfill a need we had at DreamBank: how to manage incoming applicants and resumes against available job positions and I was pretty sure it would go.

My one-weekend project became a two weekend project and a bunch more time was shoe-horned in here and there for bug fixes and improvements until it was finally usable. And then we really got serious about the other launch and development got forgotten; it was released and functional and – dare I say it – even useful but neglected.

I now present to you “Applican“, a Rails-based resume/applicant/job tracker designed (as much as it has been) as an internal tool for small & medium-sized companies. For more details see the proto-documentation here.

I consider the current release to be about v0.4: it works, it’s useful, it undoubtedly has bugs, they’re may not be critical, there’s definitely functionality to be added. It’s free, it’s open-sourced, it’s available on Github and it require Rails 2.1.

(Apologies to Robert Rodriguez for stealing and bastardizing one of the best lines in cinema history).

I have a class, it acts_as_tree. I want to be able to delete nodes from the tree and heal the rift between grandparent and children, joining them as parent-child afterwards.

Seems like it should be pretty obvious but nothing I was coming up with was doing the trick. The node was always deleting yet so were its children. Then I googled upon this forum and the solution was found. When executing a before_destroy filter on a class that acts_as_tree, put the before filter first:

class VettingStage < ActiveRecord::Base
  before_destroy :extract_self_from_chain
  acts_as_tree :foreign_key => "parent_id",  rder => "name"
  ...
end

That was easy.

A short-circuit (||=) edge case by DABlog is exactly the kind of ‘thinking about code’ that I thoroughly enjoy with Ruby. As simple an exercise as “What does this do to?”

h[:x] ||= 2

leads to an interesting exploration of the difference between || and or, what’s going on behind the scenes with ||= and ultimately made me stop and really think about the situation, though in practical terms this has never been an issue I’ve encountered. It’s just fun to walk through it in irb.

If you’re into Ruby I suggest popping over to read the article through, and do work it out in irb as you go.

Now, my two cents: if I were to read that line of code as-is in a project, I’d expect to end up with an assignment in h for key :x. The reason being:

irb(main):015:0> g
NameError: undefined local variable or method `g' for main:Object
	from (irb):15
	from :0
irb(main):016:0> g ||= 3
=> 3
irb(main):017:0> g
=> 3

Intuitively those two ||= statements seem “same” to me.