This has come up a few times at our place, and we've just worked around it some other way, but I imagine in would be useful to others too. The Test Anything Protocol (TAP) is used by a fair number of test engines. It'd be nice to have this supported from a test report perspective. You can find out more starting at the Wikipedia page for it: http://en.wikipedia.org/wiki/Test_Anything_Protocol

One thing to note: the number of tests (for example, 1..87) can either appear before or after the test run. Most emit it first, but there are some that do it the other way.

Here's an example from Git's test suite

*** t0000-basic.sh ***
ok 1 - .git/objects should be empty after git init in an empty repo
ok 2 - .git/objects should have 3 subdirectories
ok 3 - success is reported like this
ok 4 - pretend we have a fully passing test suite
ok 5 - pretend we have a partially passing test suite
ok 6 - pretend we have a known breakage
ok 7 - pretend we have fixed a known breakage
ok 8 - pretend we have fixed one of two known breakages (run in sub test-lib)
ok 9 - pretend we have a pass, fail, and known breakage
ok 10 - pretend we have a mix of all possible results
ok 11 - test --verbose
ok 12 - test --verbose-only
ok 13 - GIT_SKIP_TESTS
ok 14 - GIT_SKIP_TESTS several tests
ok 15 - GIT_SKIP_TESTS sh pattern
ok 16 - --run basic
ok 17 - --run with a range
ok 18 - --run with two ranges
ok 19 - --run with a left open range
ok 20 - --run with a right open range
ok 21 - --run with basic negation
ok 22 - --run with two negations
ok 23 - --run a range and negation
ok 24 - --run range negation
ok 25 - --run include, exclude and include
ok 26 - --run include, exclude and include, comma separated
ok 27 - --run exclude and include
ok 28 - --run empty selectors
ok 29 - --run invalid range start
ok 30 - --run invalid range end
ok 31 - --run invalid selector
ok 32 - test runs if prerequisite is satisfied
ok 33 # skip unmet prerequisite causes test to be skipped (missing DONTHAVEIT)
ok 34 - test runs if prerequisites are satisfied
ok 35 # skip unmet prerequisites causes test to be skipped (missing DONTHAVEIT of HAVEIT,DONTHAVEIT)
ok 36 # skip unmet prerequisites causes test to be skipped (missing DONTHAVEIT of DONTHAVEIT,HAVEIT)
ok 37 - test runs if lazy prereq is satisfied
ok 38 # skip missing lazy prereqs skip tests (missing !LAZY_TRUE)
ok 39 - negative lazy prereqs checked
ok 40 # skip missing negative lazy prereqs will skip (missing LAZY_FALSE)
ok 41 - tests clean up after themselves
ok 42 - tests clean up even on failures
ok 43 - git update-index without --add should fail adding
ok 44 - git update-index with --add should succeed
ok 45 - writing tree out with git write-tree
ok 46 - validate object ID of a known tree
ok 47 - git update-index without --remove should fail removing
ok 48 - git update-index with --remove should be able to remove
ok 49 - git write-tree should be able to write an empty tree
ok 50 - validate object ID of a known tree
ok 51 - adding various types of objects with git update-index --add
ok 52 - showing stage with git ls-files --stage
ok 53 - validate git ls-files output for a known tree
ok 54 - writing tree out with git write-tree
ok 55 - validate object ID for a known tree
ok 56 - showing tree with git ls-tree
ok 57 - git ls-tree output for a known tree
ok 58 - showing tree with git ls-tree -r
ok 59 - git ls-tree -r output for a known tree
ok 60 - showing tree with git ls-tree -r -t
ok 61 - git ls-tree -r output for a known tree
ok 62 - writing partial tree out with git write-tree --prefix
ok 63 - validate object ID for a known tree
ok 64 - writing partial tree out with git write-tree --prefix
ok 65 - validate object ID for a known tree
ok 66 - put invalid objects into the index
ok 67 - writing this tree without --missing-ok
ok 68 - writing this tree with --missing-ok
ok 69 - git read-tree followed by write-tree should be idempotent
ok 70 - validate git diff-files output for a know cache/work tree state
ok 71 - git update-index --refresh should succeed
ok 72 - no diff after checkout and git update-index --refresh
ok 73 - git commit-tree records the correct tree in a commit
ok 74 - git commit-tree records the correct parent in a commit
ok 75 - git commit-tree omits duplicated parent in a commit
ok 76 - update-index D/F conflict
ok 77 - very long name in the index handled sanely
# passed all 77 test(s)
1..77

A run containing a failed test might look like:

*** t0000-basic.sh ***
ok 1 - .git/objects should be empty after git init in an empty repo
ok 2 - .git/objects should have 3 subdirectories
not ok 3 - success is reported like this
#
# fail
#
ok 4 - pretend we have a fully passing test suite
ok 5 - pretend we have a partially passing test suite
ok 6 - pretend we have a known breakage
ok 7 - pretend we have fixed a known breakage
ok 8 - pretend we have fixed one of two known breakages (run in sub test-lib)
ok 9 - pretend we have a pass, fail, and known breakage
ok 10 - pretend we have a mix of all possible results
ok 11 - test --verbose
ok 12 - test --verbose-only
ok 13 - GIT_SKIP_TESTS
ok 14 - GIT_SKIP_TESTS several tests
ok 15 - GIT_SKIP_TESTS sh pattern
ok 16 - --run basic
ok 17 - --run with a range
ok 18 - --run with two ranges
ok 19 - --run with a left open range
ok 20 - --run with a right open range
ok 21 - --run with basic negation
ok 22 - --run with two negations
ok 23 - --run a range and negation
ok 24 - --run range negation
ok 25 - --run include, exclude and include
ok 26 - --run include, exclude and include, comma separated
ok 27 - --run exclude and include
ok 28 - --run empty selectors
ok 29 - --run invalid range start
ok 30 - --run invalid range end
ok 31 - --run invalid selector
ok 32 - test runs if prerequisite is satisfied
ok 33 # skip unmet prerequisite causes test to be skipped (missing DONTHAVEIT)
ok 34 - test runs if prerequisites are satisfied
ok 35 # skip unmet prerequisites causes test to be skipped (missing DONTHAVEIT of HAVEIT,DONTHAVEIT)
ok 36 # skip unmet prerequisites causes test to be skipped (missing DONTHAVEIT of DONTHAVEIT,HAVEIT)
ok 37 - test runs if lazy prereq is satisfied
ok 38 # skip missing lazy prereqs skip tests (missing !LAZY_TRUE)
ok 39 - negative lazy prereqs checked
ok 40 # skip missing negative lazy prereqs will skip (missing LAZY_FALSE)
ok 41 - tests clean up after themselves
ok 42 - tests clean up even on failures
ok 43 - git update-index without --add should fail adding
ok 44 - git update-index with --add should succeed
ok 45 - writing tree out with git write-tree
ok 46 - validate object ID of a known tree
ok 47 - git update-index without --remove should fail removing
ok 48 - git update-index with --remove should be able to remove
ok 49 - git write-tree should be able to write an empty tree
ok 50 - validate object ID of a known tree
ok 51 - adding various types of objects with git update-index --add
ok 52 - showing stage with git ls-files --stage
ok 53 - validate git ls-files output for a known tree
ok 54 - writing tree out with git write-tree
ok 55 - validate object ID for a known tree
ok 56 - showing tree with git ls-tree
ok 57 - git ls-tree output for a known tree
ok 58 - showing tree with git ls-tree -r
ok 59 - git ls-tree -r output for a known tree
ok 60 - showing tree with git ls-tree -r -t
ok 61 - git ls-tree -r output for a known tree
ok 62 - writing partial tree out with git write-tree --prefix
ok 63 - validate object ID for a known tree
ok 64 - writing partial tree out with git write-tree --prefix
ok 65 - validate object ID for a known tree
ok 66 - put invalid objects into the index
ok 67 - writing this tree without --missing-ok
ok 68 - writing this tree with --missing-ok
ok 69 - git read-tree followed by write-tree should be idempotent
ok 70 - validate git diff-files output for a know cache/work tree state
ok 71 - git update-index --refresh should succeed
ok 72 - no diff after checkout and git update-index --refresh
ok 73 - git commit-tree records the correct tree in a commit
ok 74 - git commit-tree records the correct parent in a commit
ok 75 - git commit-tree omits duplicated parent in a commit
ok 76 - update-index D/F conflict
ok 77 - very long name in the index handled sanely
# failed 1 among 77 test(s)
1..77

You can see that the diagnostic usually comes after the failed test. The numbers at the end (1..77) represents the number of tests in the suite and could preceed the run. Multiple suites may be executed as part of a single test run. IIRC, the "failed 1 among 77 test(s)" is not required, only the "ok" and "not ok" for each test.

The standards for the test anything protocol can be found here: https://testanything.org/