2012/4/21 Per Bothner
<address@hidden>
I'm going to look at the patch in detail, but before I do, a question:
Why is your srfi-64.scm.gz "for Guile, Chicken and Gambit" so very
different from the reference implementation with your patch applied?
I just mean that I had tested my srfi-64.scm on Guile, Chicken and Gambit.
srfi-64.scm pass the srfi-64-test.scm on Guile and Chicken.
There is only one fail on Gambit and I suspect dynamic-wind bug of Gambit is cause of that.
The patched testing.scm pass the srfi-64-test.scm on Guile.
But I'm not sure it to pass the srfi-64-test.scm on Chicken or Gambit.
You mention "the order and style of definitions". Could you be more
specific? I can certainly re-order definitions if that will help,
but I'd like to understand why. And I'd prefer to re-order them myself.
For example: "macro x needs to be moved before function y because of z".
For on Guile, test-with-runner should be defined before test-apply because test-apply use test-with-runner.
I saw compile errors when I compiled my srfi-64.scm on Chicken.
Of course, srfi-64.scm had passed the srfi-64-test.scm on Guile but can't be compiled then.
So, I re-ordered definitions in srfi-64.scm and make it pass.
And what is the problem with the "style"? A module issue?
No, style is not a problem; style is just style.
Humm... like these;
test-result-ref is a procedure in srfi-64.scm but a macro in testing.scm.
test-match-any, test-match-all, test-skip, test-expect-fail and %test-should-execute are defined #f and set! later.
test-on-final-simple has return value.
and so on...