Along with this, I’ve seen a shift in applications being rejected on technical grounds to applications being rejected on the basis of real or perceived Terms of Use violations or other quasi legal grounds.

This includes rejections for:

• “Scraping” Content from 3rd party sites
• Non official use of a 3rd party API
• Concerns over handling of user data

Now, while there is no checkbox for ‘I cross my heart and hope I’m not violating any treaties of the land’ during the app submission process there is a ‘Review Notes’ section where you can list any steps you’ve taken to remaining on the right side of the law.

For my apps, I’ve started including all of the following information (when relevant):

1. APIs used, along with links to API terms of service, and just a statement such as: “This application conforms to the rules and regulations for API usage at NAME OF SERVICE.”
2. Links to naming and or branding guidelines for API usage. For instance, Netflix requires that applications using their API have names in the form of “X for Netflix” not “Netflix X”.
3. Name and email contact if there are questions / concerns about usage. In particular, I’ve been involved in some projects where the application we were producing was the ‘Official’ application, but it was being published under our company name in association with another provider. So, I included the contact information for the marketing manager at the ‘official’ company.
4. If images, icon sets, etc. are included in the application that may have been used in other applications, I’ll put a quick note regarding where and how they were licensed.
5. If the application interacts with an external site that the reviewer may not be familiar with I would include a summary of that site, including things that you might consider obvious.

   An example: I’ve heard of a couple different applications for interacting with Hacker News being rejected for various versions of: “This seems like it might be shady.”

   If I was submitting such an application, I would definitely put in the review notes a quick paragraph along the lines of: “Despite the name ‘Hacker News’ is not actually about breaking into computers of violating encryption policies, but is a social news site for people interested in startups, programming and interesting technical feats.”

6. Lastly, the app store reviewers are poring over hundreds if not thousands of applications a week. It costs you nothing to be polite and human towards them.

   As a reviewer which review notes opening would put you in a more positive frame of mind about an application:

   Thanks for reviewing my application, to get started here’s a username and password for you to login and use:

   username: appletester

   password: 1happypuppy

   or

   username: apple
   
   password: password
   

   C’mon, who doesn’t like a happy puppy.

Lastly, here is the (lightly redacted) text of the Review Notes section for Flickscan – Barcode Scanning for Netflix, which was approved first try.

For a long time I’ve been itching to try and use the ruby classifier gem for something cool and I figured that this was my chance. I decided to create a quick one off website that would let people enter a post title and see if it matched (lexically) the types of things that were previously featured on BoingBoing.

The downloaded JSON representation of the data is 138.4 MB uncompressed. I’d originally considered trying to use the body text of each post as additional fodder for the comparisons, but as I didn’t have a lot of time to work on this settled for just the post titles.

To extract the titles from the JSON file I used the following script (which could be easily adapted by anyone wanting to pull other datapoints).

Now, I had a bit of a dilema. I had been planning on using the classifier gem to do a Naive Bayesian classification of the data, but the Bayes classification relies upon training data for each outcome that you want. I had data that could be used to train the classifier on what a BoingBoing post should look like, but the data dump didn’t include the surely hundreds of thousands of submissions that they received that were completely worthless.

I needed a source of post titles that would closely match the types of submissions that BoingBoing receives and settled on scraping Reddit/New for them. They are a similar type of submission and as they mostly haven’t been voted on they’re objectively pretty low quality; just the type of thing to be rejected by discriminating BoingBoing editors.

Reddit has a well documented API that’s kinda sorta RESTlike, an example: querying http://www.reddit.com/new/.json returns JSON (as you would expect). However, querying http://www.reddit.com/new/.xml returns a DIFFERENT representation of the data (it’s an RSS Feed instead of a custom XML format).

I actually ended up being rather thankful for the RSS format as sorting out the JSON data (which was nested somewhat oddly within itself IMO) was becoming a real chore. So XML + Nokogiri to the rescue and I came up with this script to get a second “bad” set of training data.

With both sets of data I could now run the classifier, training it what were BoingBoing style posts and what weren’t.

Of note is the use of the Madeleine gem which persists the training results (otherwise you would have to re-run the training every request to do a classification). I was somewhat shocked how well this worked: I ended up checking the serialized classifier into source control and just using it directly in production – no need to retrain!.

The above script was then modified into two methods with a “Classy”, non ActiveRecord backed model within the Rails application.

The final project is up at: http://www.boingable.com if you want to check it out.

Called Comfy-Helvetica, the stylesheet makes the Hacker News site look a little cleaner and more readable (at least to me) – comparison screenshot below:

godDLL published the user stylesheet as for Safari, but I’ve since moved onto Chrome and hadn’t previously had the desire to use a custom per-site user stylesheet, so I wasn’t quite sure how.

Here are the steps if anyone else needs them:

1. Install the Chrome Extension Stylish
2. From within the Chrome Menu’s click Window > Extensions
3. Click ‘Options’ within the Stylish extension listing
4. Click ‘Add New Style’
5. Enter ‘ComfyHelvetica’ in the name field and check ‘Enabled’
6. In the ‘Code’ block copy and paste the stylesheet from https://comfy-helvetica.jottit.com/user-css
7. Click the ‘Specify’ button and switch the drop down to ‘URLs on the domain’ and enter ‘news.ycombinator.com’
8. Click ‘Save’ and Enjoy

Your finished config should look like:

This is trickier than it sounds as WordPress on Nginx has some very specific setup requirements in order for pretty URLs to work. There were also lots of tutorials online about how to setup WordPress+Nginx and how to setup Rails+Nginx, but nothing with both so I figured I would document it here in the case that it helps someone else.

1. PREREQUISITES

To keep yourself from going crazy, before attempting this, make sure that you have the following installed and working:

• Nginx + Phusion Passenger
• WordPress + FastCGI
• Your Rails app working in isolation

2. SETUP DIRECTORIES

Lets assume that you have two separate site structures setup parallel to each other.

/var/www/wordpress # <--- contains php wordpress site
/var/www/railsapp # <--- contains Ruby on Rails application


3. CREATE SYMLINK

Create a symlink from the WordPress site to the Rails application public directory. It is very important that the symlink points to the "/public" subdirectory of the Rails application.

ln -s /var/www/wordpress/rails /var/www/railsapp/public


4: ADD RAILS DIRECTIVES

We need to tell Nginx to handle our Rails application separate from the PHP+WordPress setup, so it need its own location block. Nginx will check for requests coming in that start with "/rails" and apply the enclosed rules to them.


location /rails {
passenger_enabled on;
passenger_base_uri /rails; # This is referencing the symlink we setup in Step One.
}


5. PUTTING IT ALL TOGETHER

Below is the entirely of the virtual server declaration for our combined WordPress and Rails site. This would be placed within the http{} section of your nginx.conf.

You may notice two additional passenger+rails directives: passenger_min_instances and passenger_pre_start, these aren't essential for anything to work, but in concert help the Rails application to quickly respond when the server first starts or after an idle timeout.

If you’re not familiar with them, they’re a type of square barcode that’s most often used to hold a URL. You scan the QR Code with your mobile phone and Boom! you’re on a webpage, no hunt and pecking a tiny keyboard required. Which is mostly why I’m so annoyed at seeing them slapped haphazardly on people’s blogs.

Putting a QR Code on your website is like writing “You are here.” on a Post It note and then sticking it to your eyelid. It’s an accurate if decidedly unhelpful gesture.

However, having a QR Code on a printed page is great because you can always easily get back to the original document (to see if there are new comments, check out what else someone wrote, etc.).

Here’s how to accomplish showing a QR Code only on a printed webpage using Google’s Chart API and some print stylesheet declarations:

There are two separate CSS declaration blocks: one for “screen” and one for “print”. In the screen declaration we’re hiding the div containing the QR Code, and on the print we’re displaying it (and hiding a bunch of other items that don’t make sense on a printed page).

After that, place the code that actually puts the QR Code in the footer (or wherever you want), making sure that the container div’s id matches that used in the CSS.

The QR Code is created by passing the URL of the current page to the Google Charts API. Because this example is for a WordPress blog, we used the “the_permalink();” function to get the current page’s URL.

If you’re reading this online you shouldn’t see a QR Code anywhere, but printed you should find one at the bottom of the page.

Doing so was tougher than it may have appeared due to the conversation taking place last June (7 months ago). Twitter’s built-in search features: http://search.twitter.com only goes back a couple weeks. Google’s search results return some better hits, but pretty fragmented.

So, lacking other options, I whipped up the following ruby script that “deep searches” a particular Twitter user’s past messages looking for a single string.

The script takes advantage of the Twitter’s API: the magic really happens on line 12 of the above Gist: where the API URL

is called. In this case, I chose to have the response returned as JSON, but because Twitter follows REST conventions getting XML back instead would have been as easy as changing “user_timeline.json” to “usertimline.xml” in the URL.

A couple things to note about this script:

One, while it uses the Twitter API, it doesn’t rely upon you actually being authenticated to it, the calls to “/statuses/user_timeline” are free in the sense that they don’t count against your Twitter action quota. That being said, Twitter will get annoyed with you and block your IP (automatically) if you cross some magic rate limit. To be respectful of their resources, I have the script sleep between calls for 3 seconds.

Sean Cribbs led a very wide ranging talk on how to manage and contribue to an open source project at last weeks Raleigh Ruby Camp.

Some things really stuck out to me:

1. The move to Github quieted the number of people who thought they were heavy handed as they could just fork whenever they wanted.

2. Plugins as a way to involve people in the project without changing the core.

3. Day job is supporting hacking

4. Building a service business around an open source application seems like a natural type of business.

Links:

• Radiant CMS
• WordPress.org
• Github
• CodeHaus
• Upstart
• Radiant on Crack
• Redcloth
• Bluecloth
• Rdiscount

[audio:http://buzzwordcompliant.s3.amazonaws.com/sean-cribbs.mp3]

Download

I had a great time last Saturday at Raleigh Ruby Camp and met a host of Ruby on Rails developers working on interesting things.

One of the discussions that occurred was lead by Mark Imbracio of 37 Signals. I’ve posted the audio below and have linked to some of the topics that came up in the discussion. The audio quality and production isn’t great (you may need to really turn your volume up), but I got quite a bit out of the discussion and you might as well.

Links

• Capistrano
• Amazon Web Services
• Logical Volume Manager
• CentOS
• Xen
• HA Proxy
• Puppet
• HowTo: Custom Mongrel Handlers

[audio:http://buzzwordcompliant.s3.amazonaws.com/mark-deployment-talk.mp3]

Download MP3

[update 10/23 - fixed broken link to AWS]

While both are great applications, they each cost around $80 US.

Recently, Sun Microsystems (makers of Solaris, big purple servers and pony tailed CEO’s) have unleashed a new virtualization product on the world: VirtualBox, and while it lacks many of the features of the other two it’s free, easy to use and works reasonably well.

Step 1: Download

Download the latest version of VirtualBox for OS X from the support Wiki.

Step 2: Install VirtualBox

Mount the Downloaded Disk Image and run the VirtualBox.mpkg file. This should place a VirtualBox.app in your /Applications Directory

Step 3: Create a Virtual Machine

Helpful wizards jump up to walk you through creating a new Virtual Machine, but in case you miss them, click the blue “New” button in the upper left of the main VirtualBox window and follow the directions.

You will need to make 2 decisions:

1. How much RAM to allocate. I found that 512 was adequate for a stock Windows XP installation. This will vary greatly with what you are attempting to use the virtual machine for, but for cross browser testing and basic usage was fine.
2. How much HardDrive space to use. The wizard will strongly encourage you to set this “dynamically”, meaning that although you may specify that the virtual machine can use up to 25 GB of space it won’t use that much right away. Instead, it will slowly increase the amount of space taken as you use the virtual machine.

Step 4: Install Windows

At the conclusion of the Virtual Machine creation wizard, you will be prompted to begin the Windows installation process. Push F8, partition the hard drive with NTFS, enter your code, choose your TimeZone and an hour of tediousness later you’ll be dumped at the Windows Desktop.

Step 5: Setup Networking

By default, the VirtualBox networking was set to Network Address Translation (NAT), where the VirtualBox pretends that it’s in its own separate network from the network that your Mac is actually running on. While this can be useful in some specialized cases, to just get up and running it’s not necessary.

1. Shutdown the Windows virtual machine
2. With the Virtual machine selected. Click “Settings” and choose “Network”.
3. Set the Attached value to “Host Interface”
4. Make sure that the correct host interface is set in the bottom of the screen. i.e. if you’re using wireless select AirPort. If you’re plugged in, select Ethernet.
5. Restart the Windows virtual machine and you should be good to go. The machine should pickup an address automatically off of your home network. *

* There is a fairly big assumption here that you’re in a place with a working DHCP server. If this is not the case: you’re in a Starbucks, Don’t have Internet access, etc. you’re going to need to fiddle with the network settings.

[updated 9/30/2008] – I was up too late when I made the original post for this and needed to go back and fix some grammatical and formatting errors.