Comments on: Design verification capability, part 1 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/ Essays on the Continuous Delivery of High Quality Information Systems Mon, 22 Feb 2010 09:08:24 +0000 http://wordpress.org/?v=2.9.1 hourly 1 By: Rediscovering the Obvious http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/comment-page-1/#comment-75 Rediscovering the Obvious Thu, 02 Aug 2007 14:08:02 +0000 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/#comment-75 <strong>Must-Read post...</strong> I've been bickering with one of my peers about testability vs. simplicity and extensibility vs. sufficiency,... Must-Read post…

I’ve been bickering with one of my peers about testability vs. simplicity and extensibility vs. sufficiency,…

]]> By: Corey http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/comment-page-1/#comment-49 Corey Sun, 27 May 2007 15:47:25 +0000 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/#comment-49 Ad hoc testing could be more efficient if the cost of correcting deployed defects is lower than the cost of preventing them. Ad hoc testing could be more efficient if the cost of correcting deployed defects is lower than the cost of preventing them.

]]> By: Andrew http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/comment-page-1/#comment-48 Andrew Sun, 27 May 2007 14:44:13 +0000 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/#comment-48 Hi Corey! Yeah, and from the theory "post hoc" is better than "ad hoc", it's true. But the point is that in practice "ad hoc" could be more efficient than "post hoc", agree? (a) not every possible case could be covered with "post hoc" approach (b) not all the code in a module needs to be verified, there might be lots of easy and straightforward methods that wouldn't benefit that much from unit testing Or, maybe it's better to read not "ad hoc unit tests" but something like "automated smoke tests"? But now it looks to me that in this case it is a spoke from another ladder.. Hi Corey! Yeah, and from the theory “post hoc” is better than “ad hoc”, it’s true.

But the point is that in practice “ad hoc” could be more efficient than “post hoc”, agree?

(a) not every possible case could be covered with “post hoc” approach
(b) not all the code in a module needs to be verified, there might be lots of easy and straightforward methods that wouldn’t benefit that much from unit testing

Or, maybe it’s better to read not “ad hoc unit tests” but something like “automated smoke tests”? But now it looks to me that in this case it is a spoke from another ladder..

]]> By: Corey http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/comment-page-1/#comment-45 Corey Sat, 26 May 2007 23:36:26 +0000 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/#comment-45 By post hoc, I mean: After you write any new production code, you then write a corresponding unit test for that code. In this case, your development workflow probably requires that the unit test be checked in and reviewed along with the production code. Code coverage should be fairly high, but probably less than 100%. By ad hoc, I mean: Unit tests are written, but not in any consistent or complete fashion. Maybe developers write the tests for cases they consider sensitive or troublesome. Maybe test developers write unit tests independently from production developers. Either way, code coverage is probably significantly lower than 100%. By post hoc, I mean: After you write any new production code, you then write a corresponding unit test for that code. In this case, your development workflow probably requires that the unit test be checked in and reviewed along with the production code. Code coverage should be fairly high, but probably less than 100%.

By ad hoc, I mean: Unit tests are written, but not in any consistent or complete fashion. Maybe developers write the tests for cases they consider sensitive or troublesome. Maybe test developers write unit tests independently from production developers. Either way, code coverage is probably significantly lower than 100%.

]]> By: Andrew http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/comment-page-1/#comment-44 Andrew Sat, 26 May 2007 20:42:12 +0000 http://leansoftwareengineering.com/2007/05/22/design-verification-capability-part-1/#comment-44 Could you please tell what is the difference between "ad hoc unit test" and "post hoc unit test"? Could you please tell what is the difference between “ad hoc unit test” and “post hoc unit test”?

]]>