Numbering the tests made it impossible to extend the tests in logical order
without causing regtest output differences in unrelated output. Failing
tests are now instead flagged using warnings, thus giving accurate
pointers to the problematic line numbers instead of numbers without
conclusive source info. Passing tests with verbose output don't contain
any line number info. This is a prerequisite for extending the test suite
without causing false positives.
(make-column-markup (string-split str #\NewLine)))
test =
(make-column-markup (string-split str #\NewLine)))
test =
-#(let ((test-number 0))
- (define-void-function (parser location result-info strings)
- ((string? "BUG") pair?)
- (let ((input (car strings))
- (output (cdr strings)))
- (set! test-number (1+ test-number))
- (if (not (equal? input output))
- (ly:progress "Test ~a unequal: ~a. \nin = ~a\nout = ~a\n"
- test-number
- result-info
+#(define-void-function (parser location harmless strings)
+ ((string?) pair?)
+ (let ((input (car strings))
+ (output (cdr strings))
+ (result-info (or harmless "BUG")))
+ (if (not (equal? input output))
+ (if harmless
+ (ly:progress "Test unequal: ~a.\nin = ~a\nout = ~a\n"
+ harmless input output)
+ (ly:input-warning location "Test unequal: BUG.\nin = ~a\nout = ~a\n"