Comments on: A small example of the hidden dangers of dynamically typed programming languages http://pinderkent.blogsavy.com/archives/157/ Archive from the trenches of the IT World Blog. Mon, 15 Nov 2010 03:47:19 +0000 hourly 1 http://wordpress.org/?v=3.0.3 By: Tom http://pinderkent.blogsavy.com/archives/157/comment-page-1/#comment-42 Tom Wed, 10 Jun 2009 12:28:10 +0000 http://pinderkent.blogsavy.com/?p=24#comment-42 I think it's certainly true that static typing can catch some errors much quicker than those errors would be found in a dynamically typed language, and that's a perk. However, in my experience, if you are using a dynamically typed language but you have really good unit tests then type errors are usually the first kind of errors that your unit tests find. Also, even in a statically typed language if you don't have good unit tests (or you haven't done very thorough manual testing) then the likelihood is that your code is broken. In your example above, if your testing hasn't executed a particular branch in your code then you don't really know whether that branch works or not. That's just as true in Haskell as it is in Python. In response to the previous article, in my experience programmers in dynamic languages don't need to write additional tests to check types. Tests that check functionality are sufficient, after all if the program crashes with a type error then it isn't functioning correctly :) I think it’s certainly true that static typing can catch some errors much quicker than those errors would be found in a dynamically typed language, and that’s a perk. However, in my experience, if you are using a dynamically typed language but you have really good unit tests then type errors are usually the first kind of errors that your unit tests find.

Also, even in a statically typed language if you don’t have good unit tests (or you haven’t done very thorough manual testing) then the likelihood is that your code is broken. In your example above, if your testing hasn’t executed a particular branch in your code then you don’t really know whether that branch works or not. That’s just as true in Haskell as it is in Python.

In response to the previous article, in my experience programmers in dynamic languages don’t need to write additional tests to check types. Tests that check functionality are sufficient, after all if the program crashes with a type error then it isn’t functioning correctly :)

]]>