This has been driving me crazy all evening. It turns out when you use Homebrew, the installation directories for OpenSSL and Readline libraries are different. Fortunately, there’s a compilation override as mentioned on the ruby-build Issues page:
RUBY_CONFIGURE_OPTS=”–with-openssl-dir=`brew –prefix openssl` –with-readline-dir=`brew –prefix readline`” rbenv install 2.0.0-p0
You can of course change which Ruby version to install, like 1.9.3-p392.
I have literally been bashing my head into the wall this past week trying to figure out how to get RSpec view tests to pass when using Devise. I finally figured it out after a lot of trial and error because there is a particular combination of things you have to have in place to make things work.
First, the what: I added some navigation to a menu bar that only shows up if the user is logged in. It was just:
<nav>
<% if user_logged_in? %>
<%= link_to ...blah... %>
<% end %>
</nav>
The trouble is, I kept getting this error:
Failure/Error: render ActionView::Template::Error: undefined method `authenticate' for nil:NilClass
The first problem was figuring out why Devise’s helpers weren’t being loaded. According to the Devise documentation you should create a spec/support/devise.rb file like:
RSpec.configure do |config| config.include Devise::TestHelpers, :type => :controller end
But what about running view tests? The missing line is:
RSpec.configure do |config|
config.include Devise::TestHelpers, :type => :controller
config.include Devise::TestHelpers, :type => :view
end
Great! But that only solves the missing helpers. What about getting current_user to actually return a user? I tried all sorts of things and spent days searching for the answer and it comes down to stubbing out the helpers … but which object? The answer is on the view itself:
module ApplicationRspecHelpers
def stub_user
@user = double('user')
@user.stub(:id => 100)
view.stub(:user_signed_in? => true)
view.stub(:current_user => @user)
end
end
That works exactly as I want it. So now I can write my view tests like:
require 'spec_helper'
describe 'layouts/application' do
include ApplicationRspecHelpers
describe 'authentication nav' do
context 'when authenticated' do
it 'should have Log Out' do
stub_user
render
rendered.should have_selector('nav a', text: 'Log Out')
end
end
end
end
It feels like trolling, but an article I recently read on The Atlantic Wire describes weakening iPhone demand as reported in the Wall Street Journal. (Aside: I think using the term “fan boy blogs” is just belittling.) But it feels like the real story here is that Apple’s grip on the mindshare of would-be buyers is loosening. I don’t know for sure but there could be truth in this.
A huge weakness for Apple I’ve long felt has been a solid command in web services—or any web services beyond what iTunes scantily provides. I can’t embed or link to Apple maps, link to an Apple-based social feed, or even use Apple-hosted services. What about Keynote presentations or online editing of Pages documents? iMovie or Garage Band galleries or portfolios hosted by Apple would be huge.
Google and Amazon are way out ahead here and because they now pretty much have the backbone of the internet where all the web services live they command the future. Aside from programming in Objective C, I can’t use web APIs to push and pull information, can I? I use APIs and hosted services because I can multiply my developer power by a factor only limited by my access to funds—and outside funding can make that ceiling very very high. These companies are companies I can grow into, not grow out of.
Being a leader in devices is nice, but a saturation point gets reached where everyone has the same phone or the same app and the “cool” factor dictates that a rebel band find a new niche. Is it truly the rebels that define the future? I think there is a strong pull by them at the very least. It’s not just because something is labeled “awesome” that makes an app, a service, or a product a hit. Rather, when eyeballs and creative energy is being applied to something it gets better. And that means the innovation goes there. If innovation is solely left to Apple’s employees they can be hamstrung by layers of management or company directives. Innovation in the wild is unbridled creativity that finds organic applications for technology that a corporate mindmeld can’t achieve.
Does that mean that Apple doesn’t innovate? Of course it does—my iPhone and MacBooks are among the best-built and well-designed devices I’ve ever used on a regular basis. But Apple certainly doesn’t seem to care the web. They are only providing access to it. And if you also believe that the future is in device-agnostic connected information, then you have to look beyond pretty Gorilla Glass and aluminum to see where the data is going—and the data is not going to Apple’s servers.
It needs to be more mass-market.
That is what I keep telling myself. I have been dawdling on a project I have been coding in my spare time. I threw the base of it together maybe a couple of months ago? Then a few weeks ago I started shopping it around to my friends. I saw the warts and I cringed. I saw the directions this thing could take. I seriously contemplated throwing it all into /dev/null and starting over—even up to last night.
But I just had a moment of clarity. Right now.
Just ship it.
I am laid up in bed sick and here I am on my iPhone typing away on the WordPress app and I realized that my original idea is still valid. So very much so. I still might toss the codebase away a fresh copy based mostly off it, but the core of the idea is solid. I am not going to pivot.
I felt when I came up with this idea that it wasn’t a sexy idea. After demoing it to a small number of people I got lured into the trap that it wasn’t pretty enough. That it wasn’t simple enough. That my mom wouldn’t be able to grok it. That I should pivot—is the word veer these days?
No. I am on my bed and fairly lucid and I want—nay, need—exactly what I have built right now and I know of no replacement. I still know that the direction I was going to change to is a potential future but I think my current physical condition validates my efforts.
At the end of the day I know that am the one that must be the primary use case because I saw something that wasn’t available and I know how to solve this.
I am tired of getting horrible search results when trying to understand how to conceptualize 3D transforms with CSS. So I’m going to write my own help guide:
http://seqmedia.com/grokking-css-3d-transforms/
Check it out!
The popular trick most people apparently do to remove that iPhone address bar in mobile Safari is to scroll the window up by 1 pixel:
window.scrollTo(0, 1);
My problem with this is it shifts the canvas up thereby losing 1 pixel. It turns out you can specify a fraction of a pixel and the trick works just fine still:
Note: I added a 100ms delay to it because it made it work better. I tried listening to the window “load” event it just wasn’t as reliable.
This post is mostly for me to remember how to do this. 
It turns out that getting Devise (or the built-in mailer for that matter) to send email through Amazon Web Services’ (AWS) Simple Email Service (SES) is stupid easy. Most of the work is just getting SES set up properly. After that:
1. Verify some email addresses
2. Generate your SES SMTP Credentials
3. Edit your ActionMailer settings in the production.rb file
That should do it! Send yourself a test message.
Remember: you can only send to verified email addresses while in Sandbox Mode. You must request Production Mode access to send to any unverified email address. Otherwise you will probably see the Rails log error: Net::SMTPFatalError (554 Message rejected: Email address is not verified.)
This is mostly a note to myself on how to do this.
Amazon’s Simple Email Service (SES) seems pretty easy to use but working with the DreamHost DNS management proved slightly tricky because of the large amount of steps and the order of them. Getting the DNS values right was also not completely obvious. Here’s my recipe:
1. Register a Domain
2. Change the Domain to DNS-only
3. Turn on Amazon SES
4. Verify Your Email Address
5. Update DNS custom records on DreamHost
6. Wait…!
I have been working on this thing for the past few weeks. It’s a nifty idea spawned from a need I had recently, and I figured I could package it as a service—you know, SaaS stuff. You input data and the system magically formats it for you. (No, I’m not talking about the idea I tried launching 2 years ago, but something in the same vain…)
Anyways, so over this holiday break I put in some time and crafted a basic working system to demonstrate my concept. Then I put it in front of a few friends who are in my target demographic. What I heard was this:
That’s really encouraging! Yay! Proof of concept … sort of. I also heard this:
Here I had spent time crafting a syntax I thought was relatively simple. Think of it like Wiki formatting: you type raw text in and then add some special characters and boom you get neatly-formatted data. But because this SaaS is a kind of visualization tool, people want me to be able to import all sorts of spreadsheet formats and other document types. That was exactly what I was trying to avoid.
Do I listen to my users who say they just want to basically have a version of their Office suite replicated in my service? If I go down that route I’m afraid I’ll end up being very feature incomplete. (Which, incidentally, is what happened with that other startup idea I had. There was no way for me to reach feature parity.) I was hoping that by forcing people to edit in a syntax not too distant from Wiki formatting and CSVs I could skip the problem of me having to create a GUI. I’ve learned my lesson about GUIs: never make a pretty GUI because it will always be feature deficient.
And yet, I know it’s not really the syntax that’s bothering my users. Really, it comes down to the same problem I before: users already did the work and they don’t want to do it again. I can totally see that point. They want to copy or import. They want tools to edit. I also know that is a hell of a lot of work.
Being a creator is hard. I’m trying to encourage simplicity in the face of complexity. But I also need to please the people that will carry my dreams forward. The art of all of this is finding the balance between what I think is a “good enough” experience that will get my users to their end goals, at the same time I need to make things easy enough that the amount of work they do is minimized.
You may have seen this blog post by DHH: “Rails is omakase.” This pretty much encapsulates my thinking about Rails. Rails isn’t for everyone. And it certainly isn’t for every task.
What I think Rails is good for:
If you start to do other things you might wind up just trying to monkey patch the whole system or create your own objects that mimic ActiveModel things. That’s fine, just be prepared to do a lot of work.
I like not doing a lot of work. It’s like how I used to like upgrading and building my own PCs. That time has passed and now I just want to pickup my MacBook Air and be done with it. Yeah, it’s not the fastest or has the largest storage, but it gets the job done and I don’t have to worry about anything else. I’m still a techie, but I like to focus on solving the user experience problem and not worry about the plumbing of a system. I’ll leave that to DHH and the Rails core team.
Update: The tiny change that sparked a huge conversation on Github and mislav’s post summarizing the conversation.