Disqus - Latest Comments for dchelimsky

Re: Interactive Development with Clojure.spec

David Chelimsky — Tue, 18 Oct 2016 08:28:05 -0000

Good point and, yes, you can add :kind sequential? to (s/def ::code ...).

Re: Interactive Development with Clojure.spec

David Chelimsky — Tue, 11 Oct 2016 12:09:23 -0000

The `coll-of` spec doesn't deal with order, but it does maintain order when conforming. If you hand conform a coll-of spec and a sequential type, it will return a sequential type with the same values in the same order (assuming they conform) unless you specify that you want to pour them into an unordered collection (i.e. a set). Does that answer your question?

Re: Interactive Development with Clojure.spec

David Chelimsky — Wed, 05 Oct 2016 16:06:53 -0000

Fixed - thanks.

Re: Interactive Development with Clojure.spec

David Chelimsky — Wed, 05 Oct 2016 14:19:07 -0000

Fixed the gist link - thanks for the pointer.

Yes, (rand-int (count secret)) would pass the specs for those fns, but we'd see failures in score when they produce e.g. all-matches 0 and exact-matches 1. And sure, I'm sure there's a way to modify the code so that all the existing specs continue to pass (like they always return 0), and that is a good argument for a couple of explicit example based tests.

Re: Extending Transit — Cognitect Blog

David Chelimsky — Fri, 11 Sep 2015 08:23:15 -0000

Fixed, but I went with transit instead of tranist :) Thanks for pointing it out!

Re: Clojure FizzBuzz without Conditionals - Squid's Blog

David Chelimsky — Mon, 17 Nov 2014 09:27:14 -0000

One interesting side effect of this exploration is the question "what is a conditional?" For example, DerGuteMoritz's beautifully terse and clever (in the good way!) example looks like there are no conditionals, but it uses default values for keying into a map which _could_ be viewed as shorthand for (if result result default). Same with filters in other examples, etc. Of course, my example could be viewed similarly: if the score is 0, use this fn, elsif the score is 1, etc. I'm curious to hear how other people think about what is a conditional and what is not.

Re: Clojure FizzBuzz without Conditionals - Squid's Blog

David Chelimsky — Mon, 17 Nov 2014 09:22:17 -0000

Here's one more approach. It uses transformation functions like (at least) one of the other submissions, but selects the correct one based on a "score" of the number: https://gist.github.com/dch...

Re: Myron Marston and Andy Lindeman are RSpec's new project leads

David Chelimsky — Sun, 06 Oct 2013 04:33:47 -0000

And thank you, Brian, for your contributions back in the 1.0 days. iirc you were responsible for, among other things, the initial support for nested example groups, which was an important differentiator at the time. Hope you're doing well.

Re: Clean Clojure: Small functions - Connor Mendenhall

David Chelimsky — Fri, 20 Sep 2013 18:22:11 -0000

I think everybody writing and interested in this thread is interested in clean code, but not everybody agrees on what that means :)

Thanks for your reply.

Re: Clean Clojure: Small functions - Connor Mendenhall

David Chelimsky — Fri, 20 Sep 2013 05:10:36 -0000

re: readability: there are costs and limits to everything. Breaking the original function down into 8 functions introduces 7 new opportunities to get the names wrong (more possibilities for error). At least one of the resulting names, `less-than-sum-of-digits-squared`, can easily be misinterpreted as "sum the digits, then square the result".

re: usability, I think it's more likely that `square` would be used elsewhere than `sum-of-squared-digits`, so all things are not equal.

re: testability, we now have 8 functions that have to be tested instead of 1. In an object language we might declare some of those methods private and leave them tested through their consumers, but this example doesn't really address that issue.

re: density, I think there is a balance between too dense and too thin. The reason S_Mosciatti's implementations of `sum-of-squared-digits` appear dense to you is that they are complecting two orthogonal concepts (I win the buzzword contest!): splitting the number into its digits and then summing the squares. I think it becomes much less dense if those two ideas are separated, per my other comment this thread: http://ecmendenhall.github.... (would love to get your feedback on that example).

Lastly, the 3x longer but 3x more maintainable argument works in this example because you can see the entire program at one time at either size. I don't think that scales when you're talking about a 10k line program growing to 30k.

Re: Clean Clojure: Small functions - Connor Mendenhall

David Chelimsky — Thu, 19 Sep 2013 18:23:22 -0000

The name less-than-sum-of-digits-squared? sounds to me like "sum the digits, and then square the result". I think it would make more sense if it was aligned with the first function, e.g.


count-less-than-sum-of-squared-digits
      less-than-sum-of-squared-digits?
                sum-of-squared-digits

At that point, however, the three functions with almost the same name bugs me and I want to roll the 3rd one up into the 2nd one e.g.


(defn split-digits [n]
  (map (comp read-string str) (str n)))
(defn sum-of-squares [s]
  (reduce + (map #(* % %) s)))
(defn less-than-sum-of-squared-digits? [n]
  (< n (sum-of-squares (split-digits n))))
(defn count-less-than-sum-of-squared-digits [s]
  (count (filter less-than-sum-of-squared-digits? s)))

At this point every fn is a one-liner with names that clarify the implementations (to me). Personally, I'd stop at this point absent a specific re-use need. WDYT?

(ps - sorry about the formatting - when I put spaces between each function it added extra empty lines)

Re: The Plan for RSpec 3

David Chelimsky — Wed, 17 Jul 2013 16:29:50 -0000

FYI - rspec-1 was in a single repo, but there were a few problems. One was that Rails used plugins in those days, which meant downloading the entire rspec git repo (which had a long history, so it was big) every time you updated plugins. Another was that it made it difficult to do patch releases on one gem without doing the same for all. Having separate repos makes it easier to release patches independently.

Re: Rspec’s be_an_instance_of is pretty useful

David Chelimsky — Wed, 17 Jul 2013 15:53:53 -0000

Hi Toby - I commented on this on [Reddit](http://www.reddit.com/r/rub..., but in case you didn't see it or for anybody else who sees this post: RSpec does no such type conversion. If this is happening to you in a project that uses ActiveSupport, that is likely the source of the type conversion (see https://gist.github.com/dch....

re: `be_an_instance_of``, I agree that it is useful as an additional sanity check in this case because it solves a very specific problem you ran into, but I wouldn't recommend it generally because it creates additional noise and makes the code harder to change (consider that a decision to use Date instead of DateTime would require these examples changing).

Re: RSpec Team Changes

David Chelimsky — Mon, 08 Apr 2013 17:17:20 -0000

Congrats to Bradley, Jon, and Sam, and to Myron for doing such a great job of building the team.

Re: MiniTest [Rails 4 Countdown to 2013] | The Remarkable Labs Blog

David Chelimsky — Sat, 29 Dec 2012 19:00:28 -0000

I'm pretty sure the directory structure will remain as described in this post (which, btw, was actually inspired by ZenTest back when I was first working on rspec-rails), even if ActiveSupport::TestCase < MiniTest::Unit::TestCase instead of MiniTest::Spec.

Re: hone.heroku.com | Rubygems and the Dependency API

David Chelimsky — Tue, 23 Oct 2012 20:51:46 -0000

Worked as expected for `bundle install`, `bundle exec` and `bundle update`.

Re: RSpec's New Expectation Syntax

David Chelimsky — Mon, 09 Jul 2012 14:49:34 -0000

Yi - would you please submit that as a feature request to https://github.com/rspec/rs... Better place to have that discussion. Thx.

Re: Constant Stubbing in RSpec 2.11

David Chelimsky — Fri, 29 Jun 2012 11:28:39 -0000

Once again, Myron neglects to tell you that a new RSpec feature is _his_ work! Well done, Myron!

Re: RSpec's New Expectation Syntax

David Chelimsky — Fri, 22 Jun 2012 09:51:01 -0000

Rodrigo - have you seen https://github.com/rspec/rs... re: expect(obj).to receive(message)?

Re: RSpec's New Expectation Syntax

David Chelimsky — Sat, 16 Jun 2012 19:02:01 -0000

How objects are created has no impact here. Instead of `user.should be_active` you'd just say `expect(user).to be_active`.

Re: LOLtests show rspec 2x slower than minitest

David Chelimsky — Fri, 15 Jun 2012 18:27:12 -0000

Here's an equally useless benchmark: if you make 10k groups of 1 example each instead of 1 group of 10k examples, RSpec is actually faster, though admittedly not as "faster" as minitest/spec is in this example: https://gist.github.com/293...

Re: RSpec's New Expectation Syntax

David Chelimsky — Fri, 15 Jun 2012 10:41:27 -0000

What @myronmarston leaves out is that he developed this new feature! Thanks, Myron, for all the research and great work.

Re: Improving the integration between Capybara and RSpec

David Chelimsky — Thu, 14 Jun 2012 07:22:19 -0000

How do you envision this working? rspec-rails' example groups all include XXX::TestCase::Behavior from Rails, so I think we'd need something in Rails that provides RSpec all of the helpers it needs minus Rack::Test. Are you thinking there's nothing else we need from Rails for a Capy-specific type of group?

Re: Improving the integration between Capybara and RSpec

David Chelimsky — Tue, 12 Jun 2012 12:35:32 -0000

sabereent: I've been generally opposed to "acceptance" because I think it's understood differently by too many people. The naming in rspec-rails focuses on the things that are being spec'd (models, controllers, views, etc), so I'd be more inclined to use a name like "application" or "system" than "acceptance", which describes a process rather than a thing.

Re: Stop. Write a Learning Test. - The Code Whisperer

David Chelimsky — Sat, 10 Mar 2012 15:15:47 -0000

Complete tangent, but I prefer `specify { thing.should matcher }` over `it { thing.should matcher }`. The `it { should ...` syntax works well with an implicit `subject`, but "it thing should" is like nails on chalkboard to me :)