Clarify when goEncoderShouldMatchCppEncoder.

snappy_test.go

@@ -24,20 +24,29 @@ import (
 var download = flag.Bool("download", false, "If true, download any missing files before running benchmarks")
 
 // goEncoderShouldMatchCppEncoder is whether to test that the algorithm used by
-// Go's encoder matches byte-for-byte what the C++ snappy encoder produces.
-// There is more than one valid encoding of any given input, and there is more
-// than one good algorithm along the frontier of trading off throughput for
-// output size. Nonetheless, we presume that the C++ encoder's algorithm is a
-// good one and has been tested on a wide range of inputs, so matching that
-// exactly should mean that the Go encoder's algorithm is also good, without
-// needing to gather our own corpus of test data.
+// Go's encoder matches byte-for-byte what the C++ snappy encoder produces, on
+// this GOARCH. There is more than one valid encoding of any given input, and
+// there is more than one good algorithm along the frontier of trading off
+// throughput for output size. Nonetheless, we presume that the C++ encoder's
+// algorithm is a good one and has been tested on a wide range of inputs, so
+// matching that exactly should mean that the Go encoder's algorithm is also
+// good, without needing to gather our own corpus of test data.
 //
-// The exact algorithm used, though, is endianness-dependent, as it puns a
-// byte-pointer to a uint32-pointer to load and compare 4 bytes at a time. For
-// example, the "testdata/pi.txt.rawsnappy" file was generated by C++ code on a
-// little-endian system. The runtime package doesn't export endianness per se,
+// The exact algorithm used by the C++ code is potentially endian dependent, as
+// it puns a byte pointer to a uint32 pointer to load, hash and compare 4 bytes
+// at a time. The Go implementation is endian agnostic, in that its output is
+// the same (as little-endian C++ code), regardless of the CPU's endianness.
+//
+// Thus, when comparing Go's output to C++ output generated beforehand, such as
+// the "testdata/pi.txt.rawsnappy" file generated by C++ code on a little-
+// endian system, we can run that test regardless of the runtime.GOARCH value.
+//
+// When comparing Go's output to dynamically generated C++ output, i.e. the
+// result of fork/exec'ing a C++ program, we can run that test only on
+// little-endian systems, because the C++ output might be different on
+// big-endian systems. The runtime package doesn't export endianness per se,
 // but we can restrict this match-C++ test to common little-endian systems.
-const goEncoderShouldMatchCppEncoder = runtime.GOARCH == "386" || runtime.GOARCH == "amd64"
+const goEncoderShouldMatchCppEncoder = runtime.GOARCH == "386" || runtime.GOARCH == "amd64" || runtime.GOARCH == "arm"
 
 func TestMaxEncodedLenOfMaxBlockSize(t *testing.T) {
 	got := maxEncodedLenOfMaxBlockSize
@@ -469,9 +478,6 @@ func TestDecodeGoldenInput(t *testing.T) {
 }
 
 func TestEncodeGoldenInput(t *testing.T) {
-	if !goEncoderShouldMatchCppEncoder {
-		t.Skipf("skipping testing that the encoding is byte-for-byte identical to C++: GOARCH=%s", runtime.GOARCH)
-	}
 	src, err := ioutil.ReadFile("testdata/pi.txt")
 	if err != nil {
 		t.Fatalf("ReadFile: %v", err)