Compositionality Noel Welsh (09 Mar 2005 18:17 UTC)
Re: Compositionality Per Bothner (09 Mar 2005 18:49 UTC)
Re: Compositionality Donovan Kolbly (09 Mar 2005 20:37 UTC)

Re: Compositionality Donovan Kolbly 09 Mar 2005 20:37 UTC

On Wed, 9 Mar 2005, Per Bothner wrote:

> Noel Welsh wrote:
> > Why isn't this test framework compositional?  The spec
> > makes explicit mention of this, but doesn't justify it.

If I understand what "compositional" means in this context, which is to
say the ability to structure code by nesting, I think SRFI-64 can be used
in a compositional manner, although it supports a non-compositional usage
pattern as well.  Or do I misunderstand what we mean by "compositional"?

I didn't have too much trouble writing a suite (the meta-test suite I'm
working on) that recursively calls into SRFI-64.

Although... I suppose the meta-test suite relies on some form of dynamic
scoping.  If a global variable were used to back `test-runner-current',
the implementation would have to stack it's state.  Should the spec should
just clarify that an implementation is required to do so, or would that
be impossible for some implementations?  As long as you have dynamic-wind,
you can stack a global...

--
-- Donovan Kolbly                    (  xxxxxx@rscheme.org
				     (  http://www.rscheme.org/~donovan/