Re: Length of unbounded_string.

[Georg Bauhaus <rm.dash-bauhaus@futureapps.de>]

On 27.10.11 14:17, Dmitry A. Kazakov wrote:

>> A natural extension of this phenomenon into the workplace seems to
>> be writing tests.
> 
> And the reasonable one is static checks.

Static checks are insufficient in my real world, since,
as you say, a complement is needed. This is because
my world escapes being static, among other things totally
out of control.

>> *) test cases increase the likelihood of isolated pieces of
>>    software that can be forward to help desks
> 
> What is "the likelihood of isolated pieces"?

OK, bad phrasing. Parts of a program that can be run in
isolation from the rest of the program, to see some effect,
 such that no other part of the program will be called
or referenced.

>> *) writing tests means documenting and verifying assumptions.
> 
> It does not.
> 
> A proper statement should sound like: the software must be developed in a
> way that would make it testable.

OK, but testable is another void. (What is the meaning of
"testable" without invoking more definitions, recursively,
to the point of reaching epistemological mud, remembering that
this is about the *process* of *writing* tests, not about boring
formal properties of tests?)

Example: The process of writing tests results in statements that
express the expected result of an operation.
They are thus documenting some of the (intended) properties of
a type.

     thing : T1;
     assert_equals (op (thing, data), T2'(...));

Since the type system cannot express all of these assumptions,
there is more information in these statements than in the type
system alone.  Since "formal methods" imply restrictions, some
test cases are not within reach of "formal methods".

I had not intended to say that writing tests is a substitute
for writing documentation.

> This has little to do with either documentation or understanding of a
> *given* implementation, just a separate issue.

Interesting, but I had not meant to say something about existing
software that wasn't developed with testing all along driving
the writing process.