cmd/go/internal/modfetch: TestCodeRepo failures due to external repo changes

[ALTree]

$ gotip version
go version devel +fe562cebf1 Sun Nov 18 16:13:13 2018 +0000 linux/amd64

To reproduce:

$ cd src/cmd/go/internal/modfetch

$ gotip test -run=TestCodeRepo
--- FAIL: TestCodeRepo (58.16s)
    --- FAIL: TestCodeRepo/gopkg.in_yaml.v2/v2 (3.32s)
        coderepo_test.go:368: info.Version = "v2.2.2-0.20181115110504-51d6538a90f8", want "v2.2.1"
        coderepo_test.go:371: info.Name = "51d6538a90f86fe93ac480b35f37b2be17fef232", want "5420a8b6744d3b0345ab293f6fcba19c978f1183"
        coderepo_test.go:374: info.Short = "51d6538a90f8", want "5420a8b6744d"
        coderepo_test.go:377: info.Time = 2018-11-15 11:05:04 +0000 UTC, want 2018-03-28 19:50:20 +0000 UTC
FAIL
exit status 1
FAIL	cmd/go/internal/modfetch	60.812s

Spotted on the longtest builder.

[josharian]

Looks like what happened is that github.com/go-yaml/yaml added a new comment to the v2 branch: https://github.com/go-yaml/yaml/commits/v2. (Observe that previous tip was
go-yaml/yaml@5420a8b, and that new tip is
go-yaml/yaml@51d6538, which correspond to the want/have commits in the test failure.)

cc @bcmills

[andybons]

Why are we running tests against repos that are active and not controlled by us? Seems that we should be hitting repos that we control and are pre-baked.

@bradfitz

[andybons]

Adding @rsc since he wrote the original code.

[dmitshur]

Related issue #27692. It's another instance of the cmd/go tests relying on an external live repository.

[gopherbot]

Change https://golang.org/cl/153460 mentions this issue: cmd/go/internal/modfetch: update TestCodeRepo for gopkg.in/yaml.v2 v2.2.2