[TIP] Fwd: Robot Framework 2.0 is available!

Stefano Masini stefano.masini at pragma2000.com
Wed Jun 25 09:44:46 PDT 2008


On Wed, Jun 25, 2008 at 5:26 PM, Pekka Laukkanen <peke at iki.fi> wrote:
> [snip]
> The steps in
> these test cases can be very high-level, for example "Administrator
> logs into the system and there are no users in the database", and they
> tend to read better in natural language especially if they have
> punctuations or other special characters.
> [snip]
> Since lowest level
> keywords map directly to methods in library modules or classes, it is
> easy for more technical testers to start writing test libraries using
> a real programming language.

This is an interesting point. Can you elaborate a bit?

Let me see if I understood, please correct me if I'm wrong: in other
words the original customer representative actually types in
"Administrator logs into the system and there are no users in the
database" in the "TestCases" table. This piece of text is located in a
cell of the column "Action" and therefore it is called a Keyword.
This Keyword has to be properly defined in a section of the table
"Keywords", where it may break down into a sequence of other lower
level keywords, that in turn may be defined in subsequent sections of
the same table, or can ultimately be defined in python.

If this is the case, what puzzles me is who is going to be responsible
for breaking down the definition of the original long keyword into
smaller pieces.
What prevents the inexpert customer representative from writing
something like "Administrator can see what others are doing"? This
sentence can make perfect sense in spoken language, but there's
probably no way to be broken down into smaller automatable steps,
since it doesn't identify a _specific_ action. What's missing here is
probably a healthy discussion withe the customer about what are
_specific_ examples of what the sentence should mean in different
contexts, so that an actual test can be written.

I understand that my point is more on an higher level discussion about
when and under which circumstances does it make sense to have tests
actually written by someone who's not technical. As any other
acceptance testing solution Robot Framework may or may not suit the
need, besides it being a good tool, technically speaking.

Thanks,
Stefano



More information about the testing-in-python mailing list