# C[omp]ute

Welcome to my blog, which was once a mailing list of the same name and is still generated by mail. Please reply via the "comment" links.

Always interested in offers/projects/new ideas. Eclectic experience in fields like: numerical computing; Python web; Java enterprise; functional languages; GPGPU; SQL databases; etc. Based in Santiago, Chile; telecommute worldwide. CV; email.

© 2006-2015 Andrew Cooke (site) / post authors (content).

## Simple Schema for Describing Tests

From: "andrew cooke" <andrew@...>

Date: Mon, 1 Jan 2007 21:43:24 -0300 (CLST)

I've been meaning to write some notes about tests for a while now.
Unfortunately I keep thinking of different things, and then forgetting

But my main idea, which I don't expect to change, is that there's a lot of
confusion about tests because they can be viewed in several different
ways, and terminology associated with testing is often associated with a
particular viewpoint.  Worse, sometimes the same terminology means
different things depending on the viewpoint chose.

The three viewpoints I currently have on my list (entries to this list are
some of the things I keep forgetting) are:
- Extent
- Support
- Process

The EXTENT of a test can vary from a single class to a whole system.  A
common intermediate point is to test all related code for a single layer
or language (for example, testing Java code with mocks for the database,
user input, etc).

By SUPPORT I mean the infrastructure used to perform the tests.  JUnit,
for example, or Fitnesse.  Those two show how support can be related both
to extent (JUnit was originally intended for unit tests - single classes)
and process (Fitnesse was originally intended for acceptance tests).

Tests occur within the development PROCESS (whether it's explicit or
not!).  This is related to the "aim" of the test - whether it is for
developers to be check their code, make contracts explicit, or integrate
classes, or for customers to decide whether the product meets their
requirements.

Let's take some examples:

Unit tests
- extent is a single class (well, sometimes a handful...)
- support is FooUnit (depending on the langauage)
- process is development (guiding implementation, making contracts explicit)

Regression tests
- extent is typically a single test, but may extend to a complex system
- support is likely FooUnit plus an automated build
- process is development/test (tickling bugs, fixing them, and making
sure they stay fixed)

Acceptance tests
- extent is usually the whole system (or as large a portion as possible)
- support may be Fitnesse and/or GUI test tools
- process is customer acceptance

In practice, of course, every project is different and should be tested
appropriately.  What I am trying to do here is not give a fixed set of
tests, but provide a framework within which we can describe "what we are
doing", so that everyone is on the same page.  My hope is that
focussing(*) extent, support and process makes that easier - rather than
say "do unit tests" we should be saying "do tests with extent X and
support Y for process Z".

Andrew

(*) Is there some way to make Firefox use British spelling when it checks?
My OED says that focussing is perfectly OK!

### Another Test Example

From: "andrew cooke" <andrew@...>

Date: Tue, 2 Jan 2007 10:38:21 -0300 (CLST)

Here's another example.  I am writing integration tests for a web
application using Spring's MVC.  Since I'm using Spring I can do this from
within Java, using the AbstractTransactionalDataSourceSpringContextTests
class, which extends JUnit.

So the extent of the classes is pretty much the whole system, the process
is integration testing (making sure everything works together), and the
technology is Spring/JUnit.

I think this is a good example, because the technology includes JUnit,
which means that someone is going to refer to these as "unit tests".
That's the kind of confusion that arises from not specifying clearly the
extent/process (and it's easy to see why it happens - when you're
surrounded by the details of programming it's easy to lose site of the
"big picture").

Andrew