[make, remake, debugging, makefile, automation, recipes, workflow]


Make is a useful tool for automating the execution of scripts and managing datasets in a complex project. However, diagnosing errors in the workflow can be challenging with make. To address this issue, the debugger remake can be used to provide extensive error information, facilitating the debugging process and helping to ensure that the project runs smoothly.

Steps to debugging

To illustrate the steps to debugging with remake we will use an example workflow with the following directory structure.

β”œβ”€β”€ Airbnb data.Rproj
β”œβ”€β”€ README.md
β”œβ”€β”€ data
β”‚   └── listings.csv
β”œβ”€β”€ gen
β”‚   β”œβ”€β”€ output
β”‚   β”‚   β”œβ”€β”€ ordinary_hosts.html
β”‚   β”‚   β”œβ”€β”€ plot_nrreviews.png
β”‚   β”‚   β”œβ”€β”€ plot_rating.png
β”‚   β”‚   β”œβ”€β”€ review_IR.html
β”‚   β”‚   β”œβ”€β”€ review_IR2.htm
β”‚   β”‚   β”œβ”€β”€ review_model.html
β”‚   β”‚   β”œβ”€β”€ super_hosts.html
β”‚   β”‚   β”œβ”€β”€ tobit1.html
β”‚   β”‚   └── tobit2.html
β”‚   └── temp
β”‚       β”œβ”€β”€ cleaned_listings.csv
β”‚       └── cleaned_listings1.csv
β”œβ”€β”€ makefile
β”œβ”€β”€ report
β”‚   β”œβ”€β”€ report.Rmd
β”‚   └── report.html
└── src
    β”œβ”€β”€ analysis
    β”‚   β”œβ”€β”€ Rplots.pdf
    β”‚   β”œβ”€β”€ analysis.R
    β”‚   └── makefile
    └── data-preparation
        β”œβ”€β”€ cleaning.R
        β”œβ”€β”€ download_data.R
        └── makefile

Suppose the files cleaned_listings.csv and cleaned_listings1.csv were created after running the cleaning.R script were not saved in the temp folder. The files in the output folder are dependent on these files in the temp folder which will result in an error while executing make.

Here are the key steps we will take you through to debug using remake:

  • Install remake
  • Run remake
  • Use the remake --trace command to run the remake tool with tracing enabled for more details on how each rule in your makefile is executed.
  • Set breakpoints using the continue or break command to pause the execution of your makefile at specific points to debug.

Now, let’s get started!

Installing remake


On Debian systems, remake can be installed by running:

$ sudo apt-get install remake


On OSX systems, it can be installed from Homebrew or MacPorts.

$ brew install remake


Haven’t installed Homebrew yet? Check out our building block to set it up!


When you run make, it simply spots the error without further information on how the target needs to be updated but we might get more information by consulting the rules for the target.

Error output after running make

Now let us run remake:

Error output after running remake

This displays additional information. We get the line number inside the Makefile for target analysis (6) and the target that got us to this one.

For further traceback information, run remake --trace analysis or just remake --trace without target name.

Tracing output

The indentation in the first few lines containing file name and line numbers gives target level nesting. The target data-preparation was asked to be remade because it is a dependency of target analysis Thus, we get the dependency nesting as build and traverse the tree.

Finally, we also get details on the scripts that were run to build the target data-preparation This tells us to check the cleaning.R script for bugs.


Run remake -n to see the sequence of steps that will be run to execute the workflow without actually running it.

Entering the Debugger and Setting Breakpoints

The simple tracing method explained above is sometimes enough but we could also work with the built-in debugger which contains some useful commands for more efficient debugging.

Run remake --debugger , remake -X or remake --debugger analysis (with a specific target name, here: analysis) to enter the debugger.

Initiating debugger session

It first checks whether the Makefile itself is up to date. Now that we have initiated the debugger we can set a breakpoint and run until the dependent target. The debugger assigns a number to each breakpoint created. We could also set a breakpoint using a line number instead of a target name. Both continue and break commands can be used to set breakpoints.

For a given target, there are three main stopping points:

  • before target prerequisite checking: prereq
  • after target prerequisite checking but before running commands: run
  • after target is complete: end

Setting breakpoint

Adding run to the end of continue data-preparation causes the debugger to stop after dependency checking. This is the default option. Previously, the debugger stopped before dependency checking , as shown by the icon ->, so it lists the dependencies for the target. Here, the dependency for the target β€˜analysis’ was β€˜data-preparation’. Later, the debugger stopped after dependency checking, as shown by the .. icon rather than ->.

Furthermore, you can run break to get a list of all breakpoints. To delete breakpoints, run delete with the breakpoint number.


Click here for more information on the debugger syntax

Additionally, once you enter the debugger the backtrace command comes in handy to trace the root of the error.

Output from backtracing


Click here for a complete list of debugger commands.

Lastly, to exit the debugger session simply run quit.

Quitting debugger session


The following changes have been made to GNU Make to improve tracing and error reporting with remake:

  • The Makefile name and the line inside this file are reported when referring to a target.
  • A list of relevant targets with their locations is shown in the error message.
  • The command invocation used to run make is also shown.
  • There are useful options that allow for entering the debugger on error.

See also

Contributed by Roshini Sudhaharan