About
DecLog provides a framework for building unintrusive loggers for functions.
The Logger is applied to functions or methods as a decorator, and
automatically captures the arguments and return values, plus some optional
information about the environment. The user can also manually log values
with the log function. The Logger sends these values to a Database
which can be used interchangeably with different output formats. The
Database behaves similarly to a python dictionary, making it intuitive
to access past entries.
DecLog was originally designed to integrate with a data analysis library at a research facility in order to create an archive of results1. Existing logging libraries are geared towards long-running services by producing a flat feed of messages (differentiable only by level i.e. info/ warning/error). DecLog is aimed at recording function calls, and stores logged values in a hierarchical structure, using key-value pairs.
DecLog is available right now! Check out the installation guide.
-
The motivation for archiving results with function arguments and intermediate values, in addition to the code version, is that it makes the result entirely reproducible - a cornerstone of the scientific method. ↩