tpc_logger

Logging system for the TPC software.

The TPC Logger class provides a singleton for logging information within C++ code or in the python API. The class Logger contains an instance of spdlog::logger for both printing to the console (s_ConsoleLogger) and for logging to a file (s_FileLogger). There are four basic log levels:

Logger::info("Generic information...");
Logger::trace("Stack trace level of information...");
Logger::warn("Warning! Don't do what you're thinking of doing!");
Logger::error("ERROR! Program crashing!");

Logs which are saved to file utilize a revolving file structure in a hidden folder .logs which is created wherever the code is run. Running the pytest example generates a file called log.log whose entries look like:

[08:36:07] [info] File: GHDAYVXHMKXWSUAHDIVSSMQRG
[08:36:07] [info] File: MDDKVVRIPLFIJNWXOMHLHXECIPJLUJAPSIMDRJHDSSUAXDTMUGEWMYJNZSSISIVKRXNJLVEPZJFNFYFBEREGPAPQCSU
[08:36:07] [info] File: XGEAJZRSKZCRMWQVZABADVBZCQLMBPYKWSOVZDLFXMIUMD

Performance

Below are some benchmarks of performance within the python API. The source code that generates the plot below is in the /app/ directory of the main branch. The code runs a simple function that adds two numbers and either calls a generic logging message (tpc_logger.info("Adding x + y")) or doesn't. The two curves below show the difference between calling the logging function and not calling it (log - no log) and between asking whether to call it with a debug parameter and not calling it (debug - no log). Each of these calls happens within a for loop, and the number of times is given by N. The average call time is computed by taking the difference between the two curves and dividing by N (i.e. the number of calls). The error bars are the error on the mean as computed for ten trials for each value of N.

Installation

Some minimum requirements must be met in order to install:

Required Packages

• cmake (latest version)
• python (>= 3.7)
• pytest (for testing)

Once dependencies have been met, clone the repo along with its submodules

git clone --recurse-submodules https://github.com/UC-Davis-Machine-Learning/tpc_logger.git

or

git clone https://github.com/UC-Davis-Machine-Learning/tpc_logger.git
cd tpc_logger
git submodule init
git submodule update

To install, run the setup.py script:

python setup.py install

This will run cmake to compile the project and generate the python library. The default settings set the logger to only output to the log file, the console logger is turned off by default (CONSOLE_LOGGER=OFF). Unit tests are enabled by default, but can be turned off by swtiching BUILD_TESTS=OFF. To run the tests, either go to the "tests" directory and issue:

pytest .

or, go to the build directory (probably "build/temp.[environment]") and issue,

ctest

C++ usage

To use within a C++ class, first include the header Logger.hh. You can then invoke the log level commands:

#include "Logger.hh"

// some condition causes an error
if (condition == true)
{
    Logger::error("ERROR! generic error message ...");
}

// store/print some generic info
Logger::info("Calling function foo()");
foo();

Python

Basic usage in python is to first import tpc_logger which returns an instance of the singleton class. Then, issue one of the four log level commands:

import tpc_logger as logger
import sys

# some condition causes an error
if (condition == True):
    logger.error("ERROR! generic error message ...")
    sys.exit()

# store/print some generic info
logger.info("Calling funcion bar()")
bar()