Changeset 12924 for main/waeup.kofa/trunk/docs

Timestamp:

11 May 2015, 13:35:06 (10 years ago)

Author:

Henrik Bettermann

Message:

Introduction to testing.

File:

• main/waeup.kofa/trunk/docs/source/userdocs/testing.rst (modified) (2 diffs)

main/waeup.kofa/trunk/docs/source/userdocs/testing.rst

@@ -4 +4 @@
 **************************
 
-DocTests
+Kofa's Python code is being tested automaticallyy. The developers'
+goal is to reach 100% code coverage by Kofa's test runners, which
+means that each single line of code is passed at least one time when
+running the tests.
+
+Why testing? Testing makes sure that our code works properly under
+given sets of conditions and, only comprehensive testing ensures
+that changing or customizing the code does not break existing
+functionality. Why *automated* testing? Simply because no developer
+likes to click around the user interface to check tons of
+functionality. In Kofa more than 1100 tests, with dozens of actions
+per test, are being executed each time when the testrunner is
+started. This job can't be done manually.
+
+There are two different ways to test the code of a web application
+automatically: *unit tests* and *functional tests*. The goal of unit
+testing is to isolate each part of the program and show that the
+individual parts are correct. Functional tests of a web application
+are more wholistic, they require a running web application in the
+background with even a working (temporary) database. Functional
+tests are typically linked to a particular use case, they are
+interacting with the portal as a user would. That implies that
+functional testing has to make use of a test browser. A test browser
+does the same job as normal web browser do, except for visualizing
+the rendered html code.
+
+There are also two different ways to integrate tests, either
+functional or unit, into a Python package: *doc tests* (or doctests)
+and *Python tests*. Python test modules are a collection of
+isolatable Python test cases. A test case combines a collection of
+unit test methods which are being executed by the testrunner one
+after the other. Python test modules are automatically identified by
+means of their filenames which start with ``test_``. In contrast,
+doctests can be wrapped up in simple text files (ending with
+``.txt``), or they can be put into docstrings of the application's
+source code itself. Common to all doctests is that they are based on
+output from the standard Python interpreter shell (indicated by the
+interpreter prompt ``>>>``. The doctest runner parses all ``py`` and
+``txt`` files in our package, executes the interpreter commands
+found and compares the output against an expected value. Example::
+
+  Python's `math` module can compute the square root of 2:
+
+  >>> from math import sqrt
+  >>> sqrt(2)
+  1.4142135623730951
+
+Why wrapping tests into documentation? An objective of testdriven
+software development is also the documentation of the 'Application
+Programming Interface' (API) and, to a certain extent, providing a
+guideline to users, how to operate the portal. The first is mainly
+done in the docstrings of Python modules which present an expressive
+documentation of the main use cases of a module and its components.
+The latter is primarily done in separate ``txt`` files.
+
+When starting the development of Kofa, we relied on writing a
+coherent documentation including doctests in restructured text
+format. During the software development process the focus shifted
+from doctesting to pure Python testing with a lot of functional
+tests inside. It turned out that Python tests are easier to handle
+and more powerful. Drawback is, that these tests cannot easily be
+integrated into the Sphinx documentation project (the documentation
+which you are reading right now). However, we will list some of
+these tests and try to explain what they are good for.
+
+
+Doctests
 ========
 
@@ -74 +140 @@
 
 
-Python Tests
-============
+Pythontests
+===========