Here is a series of tutorial videos to get you started on the Eiffel Testing Framework (ETF).

ETF adds the ability to introduce user requirements into a seamless process of software construction at a point before concrete user interfaces can be specified. A user writes an input grammar (input.txt) that describes all possible user input operations. The etf command operating on the input grammar generates code that takes care of parsing input commands. The software designer can then focus on developing the business logic (the model).

ETF allows software developers to produce use cases that can be turned into acceptance tests, and that then free the developer to develop the business logic while not losing sight of the user requirements. This allows requirements to become part of a seamless development from requirements to implemented code, and allowing change even at the level of requirements.

(1) You need a file definitions.txt, for example:

system bank
type NAME = STRING 
new(name: NAME)
  -- create a new bank account
deposit(name: NAME; amount: VALUE)
  -- deposit "amount" into the account
withdraw(name: NAME; amount: VALUE)
  -- withdraw "amount" from the account
transfer(name1: NAME; name2: NAME; amount: VALUE)
  -- transfer `amount' from `name1' to `name2'

(2) You need an acceptance test file, e.g.. at1.txt

new("Steve")
deposit("Steve", 10)
deposit("Steve", 5)
withdraw("Steve", 10)

(3) On a Prism workstation (or SEL-VM), execute the etf command to generate starter code for the bank system

mkdir src
ls

You should see

at1.txt  definitions.txt  src/

Now do:

etf -new definitions.txt src/

The above command will generate the starting code in directory src/.

Create a temporary directory in which to compile the EIFGENs (not needed on your Laptop)

mkdir /tmp/$USER

(4) To open the generated project in EiffelStudio:

estudio16.05 src/bank.ecf &

The above command will compile the project using EStudio. There is a long compile cycle at this point, so be patient. To speed up the compile cycle use the /tmp directory as follows

(Assuming the login is “student”, i.e. $USER. Use your Prism login instead of “student”.). This will ensure that the EIFGENs are located on the current workstation in its /tmp/student directory, and thus avoid having to swap the many C files and compilations over the slow network to the NFS server, and help to speed up compilation.

After compilation, if you execute the system, you get the Graphical User Interface:

(5) We would like to run acceptance tests (such as at1.txt) in batch mode from the command line. To do this, change a switch in the ROOT class:

(6) Create a symbolic link to the workbench executable

ln -s /tmp/student/EIFGENs/ehealth/W_code/bank bank.exe

(7) You can now execute the acceptance test

bank.exe -b at1.tx

to obtain

System State: default model state (0)
->new("Steve")
  System State: default model state (1)
->deposit("Steve",10)
  System State: default model state (2)
->deposit("Steve",5)
  System State: default model state (3)
->withdraw("Steve",10)
  System State: default model state (4)

ec16.05 -c_compile -project_path /tmp/$USER -config system.ecf

You can create a symbolic link to the W_code executable:

ln -s /tmp/jonathan/EIFGENs/ehealth/W_code/system system.exe

You can now run a batch test as:

./system.exe -b at1.txt

If you recompile the code then you can execute again to see the results. You can also use the EStudio IDE to set break points and debug your code:

If you use the Haskell testing script, you could automatically test and get results such as:

indigo 189 % ./etf_test.sh 
Building executable from ETF_Test.
...
=========================
Test Results: 6/6 passed.
=========================
Success: ./log/at1.expected.txt and ./log/at1.actual.txt are identical.
Success: ./log/at2.expected.txt and ./log/at2.actual.txt are identical.
Success: ./log/at10.expected.txt and ./log/at10.actual.txt are identical.
Success: ./log/at11.expected.txt and ./log/at11.actual.txt are identical.
Success: ./log/at12.expected.txt and ./log/at12.actual.txt are identical.
Success: ./log/at13.expected.txt and ./log/at13.actual.txt are identical.
=========================
Test Results: 6/6 passed.
=========================

ec16.05  -c_compile -finalize -project_path /tmp/$USER -config system.ecf

• Haskell Script to check actual vs. expected acceptance tests.
• Using EStudio debugger to step through ETF code
• Run an ETF console-app as a web-app

In ETF, the developer places the business logic in the sub-cluster model. This sub-cluster is recursive. To add a sub-cluster (e.g. operations) to model, use the operating system to create a new folder under model with the name “operations”.

model
├── etf_model.e
├── etf_model_access.e
└── operation
    └── operation.e
    

Then, in the IDE, recompile. The sub-cluster operations will automatically be there. There is no need to use the Settings tool of the IDE to add a new cluster (in fact, this will only create problems in this recursive case).

A tests folder is also automatically generated. However, this folder is not automatically part of the system. To add this folder as a cluster, use the Settings tool of the IDE as follows: