Ron, let me be clear - Lacking any sort of true artificial intellegence, I believe executable specifications are a fool's errand. I made my definitions very...
... To the extent that "executable spec" == "automated acceptance test", I consider them quite practical and very important. Ron Jeffries www.XProgramming.com ...
... To clarify: I think the term executable spec is a dumb idea. I think if we mean we use fitnesse tests (or other things like it) to explore, augment, and...
Hello, heusserm. On Sunday, May 11, 2008, at 9:34:21 PM, you ... That readers might profit from your expressing your ideas more directly? Ron Jeffries ...
... At least 2 other people in this thread have indicated that using automated acceptance tests instead of text as the target for each iteration's mission also...
... I think that tests and specs are separate concepts. But I've never seen a tool that can make specs executable. Tests are written in terms of concrete...
Has anyone done TDD with a modeling tool like MagicDraw doing Round-Trip Engineering to generate the models? Boss at work is all into the UML stuff, and if the...
I've generated "models" (really code visualisations) using reflection and graphviz. It's very easy to do and produces useful diagrams. It can also be done...
... Well, yes, but the term "requirements document" is another dumb idea. (Even worse if it's truncated, as it often is, to "requirements.") I once mentioned...
For me, an executable spec is quite a reasonable idea and one I've used regularly inconsrained, well understood domains. If you build up your programming...
... Specs being untestable, they must also be unable to be implemented? Or do we test them with a finite number of tests? And are these tests not ... examples?...
... How many points determine a straight line? Ron Jeffries www.XProgramming.com There's a difference between righteous anger and just being crabby. --Barbara...
... The latter. ... That's what I was trying to say! For example... A customer might ask for something like "our position in a loan must be included in the...
Cory, I've done this (not really what you ask, but related): sketch some high-level models in UML; pick on a functional piece; implement 'enough' using TDD;...
... Well, we definitely agree down to the penultimate paragraph above, but my point was slightly different from the last. If we take "spec" to be this thing...
... Oh, I guess I should have made it clear. I think a suite of executable requirement illustrations are the best tool I know for getting software right. But...
... <chuckle/> Ron Jeffries www.XProgramming.com The opinions expressed here /are/ necessarily those of XProgramming.com. But I might change my mind....
... I found the following technique useful in bridging the gap between the general requirement and the need to write concrete tests: In a nutshell: Identify...
I have used MagicDraw and other similar tools to do one half of "round-trip": to generate UML-style models from code. They work pretty well for this, though...
... In my working life, they tend to act more like warning signs that somebody doesn't want to figure out the specific cases where a rule applies. mgb...
... Looks rather similar to our JUnit tests, except we don't stub anything out. We use stubs and mocks in unit tests, but not for acceptance tests. We don't...
... If you replace "it does the job" with "it may do the job" - it other words, it may satisfice - or even - it's far better than nothing - George, I think we...
... This is pretty close to equivalence class partitioning, which is one of the top tools in my arsenal to take the infinite possible number of tests and...
... Nat, I think you might be interested in the work being done on Theory based testing with regards to this. Theory explorers are particularly interesting in...
... Thanks. I know it. Another tool in the same vein is QuickCheck (http://www.cs.chalmers.se/~rjmh/QuickCheck/). Haskell's semantics and decent type system...