srfi/231 and srfi/179 tests don't follow the standard

[feeley]

In general, tests should not produce output to the terminal. The tests of srfi/231 and srfi/179 are particularly noisy and this makes it hard to check that a make test-modules does not fail a test (for another module). Perhaps a verbose? global variable should be added, initialized to #f and guarding all terminal output. That way it would be easy to activate the original behavior if needed.

Also, the srfi/231 and srfi/179 don't currently use the standard _test or srfi/64 modules, so the failures are not reported the same way, which makes finding failures harder with shell scripts, grep or other. Compare these:

$ gsi/gsi -:= srfi/4/test
*** all tests passed out of a total of 1330 tests
$ gsi/gsi -:= _test/quiet srfi/4/test
$ echo $?
0
$ gsi/gsi -:= -e '(define u8vector-ref list)' _test/quiet srfi/4/test
"lib/srfi/4/test/test.scm"@290.1: FAILED (test-eqv 0 (u8vector-ref u8v1 0)) GOT (#u8(0 255) 0)
$ echo $?
1
$ gsi/gsi -:= srfi/231/test
...
array-block!
array-decurry!
array-assign!
Failed 0 out of 11725 total tests.

[gambiteer]

All true. (Although I find the rest of the results harder to read than the results for SRFIs 179 and 231, but the others are easier to parse with a script.)

It's on my list to look at.

I've started writing investigating a test suite for SRFI 231 to test Chibi's developing implementation. Maybe with that experience I'll be able to adapt the existing testsuites.

Brad