testing: roundDown10 sometimes rounds too far

[kr]

Run http://play.golang.org/p/kX3sgrx7c4

This calls roundDown10 (copied verbatim from package testing)
with 10001 as input.

What is the expected output?

10000

What do you see instead?

1000

Solution:

Test n > 10 should be n >= 10.

This is the least consequential bug I can ever remember reporting.

I'll submit a code review if no one beats me to it.

[minux]

Comment 1:

not a big deal, but ok, please send the CL.

Labels changed: added priority-someday, removed priority-triage.

Status changed to Accepted.

[davecheney]

Comment 2:

https://golang.org/cl/9738052 adds a test which shows the issue (commented out).

[davecheney]

Comment 3:

This issue was updated by revision 787976c.

R=golang-dev, r, minux.ma
CC=golang-dev
https://golang.org/cl/9738052

[minux]

Comment 4:

I suggest this version:
func roundDown10(n int) int {
        ret := 1
        for ret < n && ret * 10 < n {
                ret *= 10
        }
        return ret
}
ps: also this patch:
diff -r fd791d51e476 src/pkg/testing/benchmark_test.go
--- a/src/pkg/testing/benchmark_test.go Fri May 31 23:03:22 2013 +1000
+++ b/src/pkg/testing/benchmark_test.go Fri May 31 21:27:16 2013 +0800
@@ -16,16 +16,16 @@
        {10, 1},
        {11, 10},
        {100, 10},
-       //      {101, 100}, // issue #5599
+       {101, 100}, // issue #5599
        {1000, 100},
-       //      {1001, 1000}, // issue #5599
+       {1001, 1000}, // issue #5599
 }
 func TestRoundDown10(t *testing.T) {
        for _, tt := range roundDownTests {
                actual := testing.RoundDown10(tt.v)
                if tt.expected != actual {
-                       t.Errorf("roundDown10: expected %v, actual %v", tt.expected,
actual)
+                       t.Errorf("roundDown10(%v): expected %v, actual %v", tt.v,
tt.expected, actual)
                }
        }
 }
i failed to submit the comment in time for CL 9738052.

[kr]

Comment 5:

Just sent https://golang.org/cl/9915045
s/least consequential/most important/
Apologies for the typographical error.

[davecheney]

Comment 6:

This issue was closed by revision a28609d.

Status changed to Fixed.