)]}'
{
  "commit": "01b01c0d12d19b4e66a54f6e48fa8aaa545a629e",
  "tree": "48cb1873cf6e35aa00a039b511da1078a641d62b",
  "parents": [
    "b433f4c40bbcdbaa88258445116a9850b49d5dc1"
  ],
  "author": {
    "name": "Yue Chen",
    "email": "yuec@google.com",
    "time": "Fri Nov 17 15:52:46 2017 -0800"
  },
  "committer": {
    "name": "Yue Chen",
    "email": "yuec@google.com",
    "time": "Sun Nov 19 00:07:22 2017 +0000"
  },
  "message": "Make the first cdf prob \u003c CDF_PROB_TOP to avoid infinite loop\n\nThe probability training tool generates all CDF_PROB_TOP table if\nthe counts are all zero. Although it means such cases are never\nused in actual encoding, in rd cost pre-calculation, we still\nsometimes run into such cases and the token cost calcualtion\nfunction cannot handle all-CDF_PROB_TOP cdfs (Will trigger assertion\nwhen debug mode is on).\n\nChange-Id: I259620e210fea0f6e1f7848724fa006fa06da226\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "8acdc0de9f70afa8831c5773e3479d09c654593d",
      "old_mode": 33188,
      "old_path": "tools/aom_entropy_optimizer.c",
      "new_id": "3108a0dc5c38e9e03fe973ccc2343d534e98929b",
      "new_mode": 33188,
      "new_path": "tools/aom_entropy_optimizer.c"
    }
  ]
}