The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

Introduction to Using Provenance to Debug with provDebugR

June 11, 2020

provDebugR is a time-travelling debugging tool that leverages provenance and is created for use on R scripts.

provDebugR uses provenance produced by rdtLite, a provenance collection library. This provenance is stored in PROV-JSON format, which follows the W3 PROV-JSON standard.

What is provenance?

Provenance is a detailed record of an execution of a script. This includes information about the steps that were excecuted and the intermediate data values that were used and/or created.

provDebugR was created to promote the use of provenance in scientific analysis, to make data analysis more reproducible and transparent.

Why provDebugR?

A typical debugger pauses execution at pre-determined points so a user can observe the execution environment. This functionality allows them to step through the program line-by-line and watch how variables change and observe where the flow of control leads.

The main drawback about this is that the user is unable to go back to lines prior to the line they are on. If they wanted to observe the execution environment at a previous line, they must start the debugger from the beginning of execution. This can be very time-consumint and also makes it hard to debug non-deterministic scripts.

By utilising provenance, provDebugR allows the user to observe the execution environment at any point in the script’s last execution, jumping back and forth between lines, as many times as they wish. It also allows for the tracking of variables throughout the script to observe type changes and connections to other variables.

Initialisation

Version 3.5.0 or later of R is required.

provDebugR must be initialised using one of the following functions before the rest of its suite of functions becomes available.

prov.debug.run(script, ...)

prov.debug.run takes in the path to an R or Rmd script and executes the script using rdtLite to collect provenance before initialising the debugger.

library(provDebugR)
prov.debug.run("myScript.R", snapshot.size = 100)

prov.debug()

Alternatively, if rdtLite’s prov.run function was just called, the provenence stored in memory can be used directly to initialise the debugger.

library(rdtLite)
library(provDebugR)
prov.run("myScript.R", snapshot.size = 100)
prov.debug()

prov.debug.file

Lastly, the debugger may also be initialised using a PROV-JSON provenance file that was created by an earlier execution of rdtLite.

library(rdtLite)
prov.debug.file("provJsonFileName.json")

Other provDebugR Functions

After the debugger has been initialised, the rest of provDebugR’s suite of functions may be used.

Functions to help debug error and warning messages

debug.error, debug.warning and debug.type.changes

These functions provide help in finding the code that led to an error or warning message. debug.error also helps find advice on Stack Overflow for the errors encountered. debug.type.changes looks for situations where the type of a variable changes as that may be an indication of a programming problem, such as when a value that a programmer expects to be a single integer turns out to be a long vector of integers.

Functions to display variable values at different points of execution

debug.line, debug.state, debug.variable, and debug.view

These functions allow the user to see the values of variables at different points of execution in the program, without the need to add breakpoints or print statements and re-run the program. The functions vary in what they query and the format of the outpu.

Function to display the lineage of a data object

debug.lineage

This function is used to show how a data object was used, or how it was produced. It shows just the lines of code that actually contribute to the value a variable has, or that depend either directly or indirectly on the value that a variale has.

End-to-end Provenance

provDebugR is one of several tools created in the end-to-end provenance group. You can find out more about our project here.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.