Disqus - Latest Comments for mguterl

Re: Does CoffeeScript Have a Future?

Michael Guterl — Thu, 15 May 2014 09:49:48 -0000

The benefit of less syntax with CoffeeScript has always been the LEAST compelling reason to use it. The semantic advantages have always drawn me to CoffeeScript over JavaScript and it seems like with ES6 there are less advantages over CoffeeScript. I don't think I will be using CoffeeScript nearly as much as ES6 gains more traction. Just my 2 cents.

Re: 6 Ways to Remove Pain From Feature Testing in Ruby on Rails

Michael Guterl — Sat, 05 Oct 2013 11:20:01 -0000

I agree that regex's aren't the problem with Cucumber. I find that most people don't follow the prescribed best practices when writing features and it causes a lot of pain.

Joel did a great job of summarizing some of our thoughts on Cucumber here: http://gaslight.co/blog/how...

Re: Better Code Design through Pictures

Michael Guterl — Sun, 29 Sep 2013 08:46:45 -0000

This morning I stumbled upon https://www.draw.io which looks like an awesome tool for doing more of this type of work. I really enjoy whiteboarding, but I don't enjoy the artifact (a photograph). I think translating the result to something more formal using https://www.draw.io might be an interesting idea.

Re: Angular backed SVGs

Michael Guterl — Thu, 08 Aug 2013 07:16:56 -0000

Thanks for pointing out sullerandras's web sequence diagram generator. That is great!

Re: Gaslight: How We Handled the Problem of JS Date/Time Equality

Michael Guterl — Fri, 12 Jul 2013 13:27:09 -0000

Same reply to the same question from http://www.reddit.com/r/jav...

I was unaware of deepEqual. Thanks for the pointer.
I also wanted to be able to assert on just the date portion which I feel is useful.

Re: Rails Controller Accessors by Nick Gauthier on 2013-04-15

Michael Guterl — Tue, 16 Apr 2013 07:58:26 -0000

A lot of folks that I work with really love decent_exposure: https://github.com/voxdolo/...

I haven't taken enough time to dive into it yet, but I thought you might be interested.

Re: Searching fora HeroGood Examples

Michael Guterl — Sun, 27 Jan 2013 17:54:06 -0000

Awesome, I'm glad to hear that you're going to be adding better examples. I think Hero has a lot of potential for putting some standards in place for these types of tasks. Only after years of Rails have I established a convention for this myself, so I agree with your point about discipline.

You might want to check out ruote: https://github.com/jmettrau... - I'm not sure if it is complementary to what you're doing, or if it might provide any inspiration.

I certainly think a standard approach to these types of tasks will help organize applications.

Re: Searching fora HeroGood Examples

Michael Guterl — Sun, 27 Jan 2013 13:41:18 -0000

You'll just have to trust that I would have been driven to the same or similar design. :-)

I've used the structure app/jobs and app/services to achieve a similar breakdown in some applications that I've worked on. The app/jobs folder consists of classes that respond to perform (typically consumed from Resque). The app/services folder consists of classes that wrap those third party libraries.

If were to organize the code above in a Rails app, it'd look like this:

app/jobs/deliver_news_notifications.rb
app/mailers/news_notifier.rb
app/services/news_sources/hacker_news.rb
app/services/news_sources/reddit.rb
app/services/news_sources/google.rb

I appreciate the fact Hero can provide guidance for modeling these processes. However, I would have been guided to this design for the following reasons:

1. I would want to ensure (test) that each news source works and I would have done that in isolation.
2. I want to have an interface that my application uses (gather_news in this case) and not depend on some low level details (HTTP, etc.)
3. I would want to test that given certain inputs the news notifier rendered the email body correctly.
4. I would want a single place for the wiring of these components to occur (app/jobs/deliver_news_notifications.rb).

I look forward to hearing more of your thoughts.

Re: Better Support with Test Harnesses

Michael Guterl — Thu, 15 Nov 2012 16:24:00 -0000

It also appears that you could replace the implementation of TodoOnPage with something completely unrelated to a page or the web and it could still continue to function testing your domain logic.

Re: Refactoring with Hexagonal Rails

Michael Guterl — Wed, 18 Jul 2012 09:00:40 -0000

I'm currently refactoring some of my code to adopt some of the ideas presented here. I'm finding that the code is much easier to understand and modify.

I'm curious how are you organizing this structure on the filesystem? It seems like you'd want PlanChange like objects in one place and subscribers in another, but what do you actually call these objects?

Do you see any problem with avoiding PersistenceResponse and just implementing card_updated and card_update_failed directly on the controller?

You also mention that your KISSMetrics code isn't run during the test environment because PlanChange does not add the class as an observer. How are you handling the creation of PlanChange objects per environment?

Re: SOLID JavaScript: The Single Responsibility Principle

Michael Guterl — Sat, 17 Dec 2011 12:31:51 -0000

It seems to me that ProductsController is overloaded with responsibility that could belong to the ProductRepository, specifically finding the product by id. What do you think?

Re: Tidy views and beyond with Decorators

Michael Guterl — Fri, 02 Dec 2011 16:26:59 -0000

Thanks for your feedback guys, makes perfect sense.

Re: Tidy views and beyond with Decorators

Michael Guterl — Fri, 02 Dec 2011 15:41:38 -0000

I'm curious why you chose to use 2 different decorator classes and not simply use a MonkeyDecorator?

Re: Test after_commit hooks with transactional fixtures enabled

Michael Guterl — Wed, 20 Apr 2011 16:37:12 -0000

Mat, I can't tell you how many beers I owe you for Sunspot and related help and now you're writing patches for my inquiries on other mailing lists!

We went back to after_save callbacks and relied on Resque retrying the failed jobs when the record wasn't found. I look forward to adding your patch and going back to using the transaction hooks.

Thanks so much for your many open source contributions.

Re: http://blog.kiskolabs.com/post/776939029

Michael Guterl — Tue, 25 Jan 2011 13:11:56 -0000

Was anyone ever able to figure out this issue with CSS and JS?

Re: Making Statistical Social Validation Fast - rywalker.com

Michael Guterl — Tue, 28 Dec 2010 08:38:07 -0000

validates_presence_of_social_validation :on_signup => true

Re: Integration Testing Resque with Cucumber

Michael Guterl — Wed, 18 Aug 2010 08:41:09 -0000

We also use Resque and integration tests (rspec only though, no cucumber) for a fairly large application. Our approach is much simpler, we have a helper method called `do_work` that looks like this: http://gist.github.com/534574

The first line of the method is for resque-scheduler which we also use.

Re: Extending Formtastic with a sprinkle of jQuery - diminishing.org

Michael Guterl — Tue, 13 Jul 2010 12:03:49 -0000

Wouldn't that include the javascript and stylesheet each time comboselect_input is called?

Re: Extending Formtastic with a sprinkle of jQuery - diminishing.org

Michael Guterl — Tue, 29 Jun 2010 14:55:18 -0000

Thanks for mentioning this Ajescudero, this is in my original markdown content, but for some reason it's not being preserved in the conversion to HTML. I'll take a look.

Re: This blog is closed.

Michael Guterl — Fri, 14 May 2010 09:12:01 -0000

No, don't go!!!

Seriously though, thanks for all the awesome snippets, libraries, etc.

Re: paperplanes. ActiveRecord's Callbacks Ruined My Life

Michael Guterl — Sat, 08 May 2010 09:48:18 -0000

Mathias, I agree with you and your point about turning controllers into services. I've always disliked putting this logic in the controller, because the tests are tied to the HTTP request / response cycle. I much rather use the approach that James describes and I'm not all that worried about the extra files or classes if it means extra clarity.

Re: Rails Integration Testing Stack for Winners - Will Leinweber

Michael Guterl — Mon, 03 May 2010 09:05:27 -0000

You should check out the capybara-envjs driver, it is much simpler to use compared to C(e|u)lerity.

http://github.com/smparkes/...

Re: Crazy, Heretical, and Awesome: The Way I Write Rails Apps | James on Software

Michael Guterl — Sat, 27 Mar 2010 21:57:15 -0000

James, I'm curious where / how you handle the instantiation of the Service objects? Do you instantiate it in the controller?

Re: Crazy, Heretical, and Awesome: The Way I Write Rails Apps | James on Software

Michael Guterl — Sat, 27 Mar 2010 10:36:41 -0000

James, thank you. Your post and the comments have been a great read. As a developer who was first introduced to OOP through Rails, I followed the same pattern of development described by sandimetz. I always found functional (read: controller) testing to be awkward and enjoyed how much easier testing was with fat models. Until our models became so fat, it became a maintenance nightmare.

At the office, my coworkers regularly hear me ranting about the context in which model creation occurs. I've taken a half-baked approach to what you describe above in certain situations where we've felt some pain. While my approach was slightly different than yours, I like what you present better. I think we're going to use this pattern much more in order to bring some much needed flexibility to our app.

We've recently taken a liking to using Presenters; I'm curious how you think they fit alongside the Service pattern? What you've laid out so far is immensely helpful, but I agree with sandimetz that some clear guidelines would help others in adopting this pattern.

I believe the future of my Rails development will likely be using the MVCPS pattern.

Re: Blogging with Nesta - Oscar Del Ben

Michael Guterl — Sun, 21 Mar 2010 13:55:28 -0000

Thank you for this, I'm currently using it in my project. I'm curious if you see any reason we couldn't be using the rack-pygments middleware instead of this hack? After all Nesta is just a Sinatra application that is using Rack anyways.

I tried adding "require 'rack/pygments" and "use Rack::Pygments" to config.ru but it did not work. Am I missing something obvious why this wouldn't work?

I ask because I also had to hack the Page#body method in addition to FileModel#to_html in order to make all the generated html have syntax highlighting.