Trace

Trace provides a minimal interface for zero overhead program tracing in julia.

Macros

Macro versions of the logging methods (e.g., log, debug, info) provided in Memento are used to only include logging expressions in the compiled source when tracing has been enabled (Trace.enable()).

Example)

using Memento
using Trace

# Configure Memento logging as usual.
logger = Memento.config("debug"; fmt="[{level} | {name}]: {msg}")

# NOTE: We're enabling tracing prior to loading any of the macro
# expressions that depend on it.
Trace.enable()  # Sets `Trace.ENABLED` and adds a "trace" logging level.

# Might want to include another package that includes tracing expressions.
# using MyTracedPkg

# Log a debug message when tracing is enabled.
@debug(logger, "Something to help you track down a bug.")

NOTE: The @debug expression would be excluded at runtime if Trace.enable() was commented out.

Tracing functions

As program tracing lies somewhere between logging and profiling it is not uncommon to want to automatically log execution time and memory allocation of a function. Trace provides an @trace(logger, ex) macro for doing exactly this.

Example)

# Assuming similar code to above
...

# Set the logging level to "trace" (5) which defaults to having a
# lower priority logging than "debug" (10).
set_level(logger, "trace")


# Same behaviour as `@debug` with regard to `Trace.enable()`.
@trace(logger, rand(1000, 1000))