)]}'
{
  "commit": "e230a0bfec642600a90668d01fc30fca987ae9ec",
  "tree": "5b9f6ebf1ae958a8c8bc7d99018562ae785b834f",
  "parents": [
    "756beb5dfed42a39529807f32fc3b5938907f193"
  ],
  "author": {
    "name": "Oliver Newman",
    "email": "olivernewman@fuchsia.infra.roller.google.com",
    "time": "Wed Feb 24 16:47:42 2021 +0000"
  },
  "committer": {
    "name": "Copybara-Service",
    "email": "copybara-worker@google.com",
    "time": "Wed Feb 24 08:48:43 2021 -0800"
  },
  "message": "[roll] Roll fuchsia [testsharder] Sort tests pseudo-randomly within shards.\n\nThis is a reland of fxrev.dev/487866.\n\nThe original change ordered tests by GN label, so tests in similar areas\nof the codebase were run very close together. This broke some Scenic\ntests that do asynchronous cleanup of global state, and each assume\nclean state when starting back up (see fxb/67316).\n\nTo avoid that, sort by a hash of each test\u0027s name for a pseudo-random\nbut stable ordering.\n\nOriginal change\u0027s description:\n\u003e Revert \"[testsharder] Sort tests alphabetically within shards\"\n\u003e\n\u003e This reverts commit e3d738d0045d6a38e99c56f398463430811e54b7.\n\u003e\n\u003e Reason for revert: reordering broke some Scenic tests: ci.chromium.org/b/8854458567977869968\n\u003e\n\u003e Original change\u0027s description:\n\u003e \u003e [testsharder] Sort tests alphabetically within shards\n\u003e \u003e\n\u003e \u003e Sorting tests by duration within shards means that test duration CIPD\n\u003e \u003e package rolls can cause hard-to-debug failures in non-hermetic tests,\n\u003e \u003e and the burden of debugging those failures is on me (as the owner of the\n\u003e \u003e test-durations-roller) and build gardeners in the case of flaky failures\n\u003e \u003e that are able to get through CQ.\n\u003e \u003e\n\u003e \u003e Ordering alphabetically should all but eliminate these kinds of\n\u003e \u003e failures.\n\u003e \u003e\n\u003e \u003e See bug for full rationale.\n\u003e \u003e\n\u003e \u003e Since it drastically re-orders shards, there is a chance that this\n\u003e \u003e change itself breaks some non-hermetic tests, so I\u0027ll keep a close eye\n\u003e \u003e on things after it lands. I\u0027ll also run all internal global.try builders\n\u003e \u003e on this change make sure that it won\u0027t break internal CQ.\n\u003e \u003e\n\u003e \u003e Original-Bug: 70700\n\u003e \u003e Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/487866\n\u003e\n\u003e Original-Bug: 70700\n\u003e No-Presubmit: true\n\u003e No-Tree-Checks: true\n\u003e No-Try: true\n\u003e Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/490722\n\nOriginal-Bug: 70700\nOriginal-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/490723\nOriginal-Revision: ed80972e61a973aac6d58dbbea558cfd11f07730\nGitOrigin-RevId: 91e742843f6ebc44751df6b78c95a2265d5392df\nChange-Id: Ia10ce2498dd2c0981ce827e0f9f1aa80f502d7a4\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "5a8be96ecf5258a0bfa5eb38afdf311f39cd7f49",
      "old_mode": 33188,
      "old_path": "stem",
      "new_id": "06237f7840fc194984490f96db5b5612085e3738",
      "new_mode": 33188,
      "new_path": "stem"
    }
  ]
}