Fabian Haupt · Apr 19, 2017 2m read

Visualizing the data jungle -- Part III. Excursion to heat maps based on %SYS.MONLBL

Code coverage and performance optimization of code has come up a bunch of times already, so most of you should already be aware of the SYS.MONLBL utility.
Often a visual approach to looking at code is much more intuitive than pure numbers, which is pretty much the whole point of this article series. This time we will take a slight excursion away from python and its tools and are going to explore generating heatmaps from ^%SYS.MONLBL reports.

As a quick reminder a heatmap is just a specific visualization tool, which gives us an overview of data where colors represent a certain value. In our case the data is going to be lines of code, with the time being spent in them mapped to colors.


For details about running the line-by-line monitor, check out the documentation. In short, are going to work with the full output of an analysis as CSV file. It is a lot more useful if we actually have the sourcecode to the code we are trying to analyse. Be sure to compile your code with the k flag (keep source).

Prepare the output

As targe output we are going to use a prepared html file. This will just include the very basic layout and a little javascript function to do the final coloring.

<!doctype html>
<html class="no-js" lang="">
        <meta charset="utf-8">
        <meta http-equiv="x-ua-compatible" content="ie=edge">
        <meta name="description" content="">
        <meta name="viewport" content="width=device-width, initial-scale=1">

        <!-- Place favicon.ico and apple-touch-icon.png in the root directory -->
        <link rel="apple-touch-icon" href="apple-touch-icon.png">

        <link rel="stylesheet" href="">

        <!--<link rel="stylesheet" href="css/main.css"> -->

        table, th, td {
            border: 1px solid black;
            border-collapse: collapse;
        pre {
        th {
            text-align: left;


    <body onload="colorize()">
        <!--[if lt IE 8]>
            <p class="browserupgrade">You are using an <strong>outdated</strong> browser. Please <a href="">upgrade your browser</a> to improve your experience.</p>

<table id="data">
<tr><th>Routine</th><th>Line</th><th>Total Time</th><th>Code</th></tr>


Parsing and putting it together

Getting the relevant information out of the generated CSV and putting it into our template is being done by the following scripts:


cat $1|grep -vi totals| awk -F"," 'FNR>1 {out="<tr><td>"$1"</td>" "<td>" $2 "</td><td>" $54 "</td><td><pre>"; for(i=55;i<=NF;i++){out=out$i","}; out=substr(out, 1, length(out)-1) "</pre></td></tr>"; print out }'

./ $1 > /tmp/
sed -e '/<!--output-->/r/tmp/' template.html

Which we call like this:

./ /tmp/report.csv > heatmap.html

The final output


If you take a closer look at the colouring function in our template, you'll see I'm not using a linear mapping for the times:

    function colorize() {
        var rows=$("#data tr")
        var max=Math.max.apply(Math,rows.slice(1,rows.length).map(function(){ return this.childNodes[2].textContent}))
        for (i=1;i<rows.length;i++){
            var val=rows[i].childNodes[2].textContent;
            var c=(Math.pow(1-val/max,3))*255;
            var col=rgba(255,c,c,0.7);

I found this to work fairly well with the examples I tested, but your mileage may vary. Obviously you can increase the exponent to push it more into the red, or vice versa.


You can find all of the relevant files here

1 425
Discussion (0)0
Log in or sign up to continue