{
    "info": {
        "author": "Edison Gustavo Muenz",
        "author_email": "edisongustavo@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: Apache Software License",
            "Operating System :: Unix",
            "Programming Language :: Python",
            "Programming Language :: Python :: 2",
            "Programming Language :: Python :: 2.7",
            "Programming Language :: Python :: 3",
            "Programming Language :: Python :: 3.4",
            "Programming Language :: Python :: 3.5",
            "Programming Language :: Python :: 3.6",
            "Programming Language :: Python :: 3.7",
            "Programming Language :: Python :: Implementation :: CPython"
        ],
        "description": "A small library for publishing metrics when running a workflow using the Benchmark AI infrastructure.\n\n# Installation\n\n```\npip install benchmarkai-client-lib\n```\n\n# API\n\nThe library provides a single method:\n\n```\ndef emit(metrics: Dict[str, float])\n```\n\nPS: This library also works with Python 2, which has no type system.\n\n## Example usage\n\n```python\nfrom benchmarkai import emit\n\ndef my_fancy_benchmark():\n    import time\n    begin = time.time()\n    for epoch in range(100):\n        execute_training()\n    elapsed_time = time.time() - begin\n    emit({\"alexnet.batchsize16.latency\": elapsed_time})\n```\n\n## Performance\n\nAs the author you don't need to worry about `emit()` being slow since it does 2 things:\n\n- serializes `metrics` into JSON\n- writes this string into a file (see the internal details)\n\nSo it will be as slow as these operations can be, which are quite fast.\n\n## Accepted values\n\nThe dictionary provided by the `metrics` parameter is in the form:\n\n`\"metric-name\" => value`\n\n- `metric-name` is a `string`\n- `value` is a `float`\n\n# Running tests\n\nTo set up the environment to run the tests, the file `test-environment.yml` is provided. This conda environment file\nhas all dependencies required to run the tests.\n\n```\n$ conda env create -f test-environment.yml\n$ conda activate benchmark-ai-client-library-tests\n$ tox\n```\n\n# Internal details\n\n## Modes\n\nThe library can be used in these modes as of now:\n\n1. Stdout printing (this is the **default** mode)\n1. Benchmark AI infrastructure\n\nTo control which behavior is desired, the environment variable `BENCHMARK_AI` can be set to one of the values:\n\n- `stdout`: The **default** mode. In this configuration, the `emit()` method is equivalent to:\n    ```python\n    def emit(metrics):\n        import json\n        print(json.dumps(metrics))\n    ```\n\n- `fifo`: When running inside Benchmark AI infrastructure this is the mode used.\n\n### Fifo mode\n\nIn this mode, the `emit()` method writes JSON string to a file that was created as a \n[named pipe](https://docs.python.org/3.7/library/os.html#os.mkfifo). It does **NOT** write to `stdout`.\n\nKeep in mind that the client library does **NOT** create the *named pipe* file, this is delegated to the\n\"daemon process\", which will be running alongside the benchmark process in the Benchmark AI infrastructure.\n\nThe daemon process is responsible to:\n\n- create the *named pipe*\n- read the contents from this pipe\n- post the contents of this pipe to the metrics servers (eg.: Elastic Search, Cloudwatch, etc.)\n\nThe first time that `emit()` is called, it checks for the existence of the file and uses that. If it doesn't exist, it\nwaits for some time before giving up and throwing an exception `FifoNotCreatedInTimeError`.\n\n#### Fifo mode configurations\n\nThese environment variables control the behavior of the client library:\n\n- `BENCHMARK_AI_FIFO_FILEPATH`: The path to the file that will be created as a *named pipe*.\n                                Default value: `/tmp/benchmark-ai-fifo`\n- `BENCHMARK_AI_FIFO_MAX_WAIT_TIME`: The maximum amount of time that the client library will wait for the\n                                     *named pipe file* to be created.\n                                     Type is **float**, unit is **seconds**, default value: `10`.\n- `BENCHMARK_AI_FIFO_WAIT_TIME_STEP`: The amount of time between each check that the client library will wait before\n                                      checking for the existence of the *named pipe file* again.\n                                      Type is **float**, unit is **seconds**, default value: `0.5`.\n\n**IMPORTANT**: These environment variables are for internal usage of the Benchmark AI infrastructure. They are listed\n               here for documentation purposes only.",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/MXNetEdge/benchmark-ai",
        "keywords": "",
        "license": "Apache",
        "maintainer": "",
        "maintainer_email": "",
        "name": "benchmarkai-client-lib",
        "package_url": "https://pypi.org/project/benchmarkai-client-lib/",
        "platform": "",
        "project_url": "https://pypi.org/project/benchmarkai-client-lib/",
        "project_urls": {
            "Homepage": "https://github.com/MXNetEdge/benchmark-ai"
        },
        "release_url": "https://pypi.org/project/benchmarkai-client-lib/0.1/",
        "requires_dist": null,
        "requires_python": ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*",
        "summary": "A client library for posting metrics that uses the Benchmark AI framework.",
        "version": "0.1"
    },
    "last_serial": 4873932,
    "releases": {
        "0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "1efa1e34a87adf0aa69390e09df275b9",
                    "sha256": "7d5118375a2f81b7d8bb7bbe1e3ad7319f17e79cdcb1bdb30c41d2cd9d28c566"
                },
                "downloads": -1,
                "filename": "benchmarkai-client-lib-0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "1efa1e34a87adf0aa69390e09df275b9",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*",
                "size": 4304,
                "upload_time": "2019-02-27T12:13:09",
                "url": "https://files.pythonhosted.org/packages/93/ff/3acf083f8c8c25fe8e8043e98fc2d3fb61d1b55f23a470ac210954c4e82f/benchmarkai-client-lib-0.1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "1efa1e34a87adf0aa69390e09df275b9",
                "sha256": "7d5118375a2f81b7d8bb7bbe1e3ad7319f17e79cdcb1bdb30c41d2cd9d28c566"
            },
            "downloads": -1,
            "filename": "benchmarkai-client-lib-0.1.tar.gz",
            "has_sig": false,
            "md5_digest": "1efa1e34a87adf0aa69390e09df275b9",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": ">=2.7, !=3.0.*, !=3.1.*, !=3.2.*, !=3.3.*",
            "size": 4304,
            "upload_time": "2019-02-27T12:13:09",
            "url": "https://files.pythonhosted.org/packages/93/ff/3acf083f8c8c25fe8e8043e98fc2d3fb61d1b55f23a470ac210954c4e82f/benchmarkai-client-lib-0.1.tar.gz"
        }
    ]
}