Or a regression suite that checks that new changes does not mess up existing code.
Accordingly code offered to this site can contain unit tests, using the methods
described in this page.
But before going on, we stress that
awk.info gratefully accepts
awk contributions in any form. That is, including unit tests with code is optional.
Files
If your code is in directory yourcode then create a sub-directory yourcode/eg
Write a test in a file yourcode/eg/yourtest. Divide that test into two parts:
In the first paragraph of that file, write any tedious set up required to get the system ready for the test.
In the second, third, etc paragraph, write the code that shows the test
For example, in the following code, the real test comes after some low-level environmental set up:
# assumes
# - the LAWKER trunk has been checked out and
# - .bash_profile contains: export Lawker="$HOME/svns/lawker/fridge"
. $Lawker/lib/bash/setup
gawk -f join.awk --source '
BEGIN { split("tim tom tam",a)
print join(a,2)
}'
Write the expected output of that test case in yourcode/eg/yourtest.out
Regression Tests
The above file conventions mean that an automatic tool can run over the entire code base and perform a regression test (checking if all the tests generate
all the *.out files.
Displaying the Tests (and Output)
Another advantage of the above scheme is that you can use the tests to document your code.
To show the test case, add the following into your .awk file:
Then zip the directory yourcode (including yourcode/eg) and send it to awk.info. Once
we install those files on our site then
when awk.info displays that file, the test case trivia is hidden and the users only see the essential details.
For an example of this, see http://awk.info/?gawk/array/join.awk.
