Most development activity tends to be located in relatively few modules`. A Hotspot analysis
helps you identify those modules where you spend most of your development time.
Combined with CodeScene’s code health perspective, you can quickly prioritize technical debt or other code
quality issues:
Low code health in a development hotspot is expensive. Prioritize improvements here.
Low code health in stable parts of the codebase, i.e. non-hotspots, has lower priority.
Hotspots are an excellent starting point if you want to find your productivity bottlenecks in code.
A large codebase will contain many different hotspots. Often, there
will also be clusters of hotspots. Clusters may indicate that an
entire component or package is undergoing heavy changes.
The visualizations on theis page are available from the analysis menu
at “Code > Hotspots” or by clicking on “View Hotspots” on the “Code
Health” tab of the main analysis dashboard.
The Hotspots map in CodeScene is a flexible analytic tool that can be adapted to your specific needs.
This kind of visualization is used in several other places inside
CodeScene, so it’s worth learning how to use these maps, which we refer to sometimes as System Maps.
A Hotspot Map has several use cases and also serves multiple audiences
like developers and testers:
Developers use hotspots to identify maintenance problems.
Complicated code that we have to work with often is no fun. The
hotspots give you information on where those parts are. Use that
information to prioritize re-designs.
Hotspots points to code review candidates. At CodeScene we’re big fans
of code reviews. Code reviews are also an expensive and manual
process so we want to make sure it’s time well invested. In this
case, use the hotspots map to identify your code review candidates.
Hotspots are input to exploratory tests. A Hotspot Map is an
excellent way for a skilled tester to identify parts of the codebase
that seem unstable with lots of development activity. Use that
information to select your starting points and focus areas for
exploratory tests.
The hotspots map is interactive and hierarchical. Each large blue circle represents a folder in
the codebase. That means you can zoom in and out to the level of detail you’re interested in.
Fig. 20 Hotspots show you the activity in your codebase.¶
To zoom in and out, you can
click on a directory or file circle;
scroll with your mouse wheel;
use pinch-to-zoom gestures;
click on the breadcrumbs at the top of the view
click on subfolders in the “Content” tab of the node details in the sidebar.
You can also “grab” the system map with your cursor to pan left, right, up or down.
When you click on a circle, you’ll see information about that file or
directory in the sidebar, as illustrated in
Fig. 21.
Fig. 21 Click on a Hotspot to access the context menu.¶
The Hotspot tab is our recommended starting point. With the Combined Aspects feature, it helps the
team building a shared mental model of what the code looks like, and where the weak and strong parts are.
The Combined Aspects are available as buttons on the default “Hotspots” tab of
the Hotspots system map. When you select a button, the corresponding metric is included in calculating the color density of each file in the map.
Fig. 24 Hotspots let you visualize large codebases by combining a code health perspective with temporal and organizational data.¶
This allows you to combine different criteria to easily identify the
most interesting hotspots.
Hotspots: This is the relevance of the findings – the priority. The metric is calculated from the development
activity in the code.
Low Code Health: This is the severity of any hotspot. The lower the code health, the more red the corresponding circle that
represents the hotspot.
Defects: Defects are calculated from a project management tool like Jira. We explain this integration in more detail below.
CodeScene caches your selected combinations in your browser, meaning you can define your own hotspot perspective based on the KPIs that are
important to you.
Visualize relationships between files with change coupling¶
CodeScene identifies files that tend to change together.
Change coupling indicates a relationship of some kind between files.
These relationships are visible in the hotspot map when you hover your cursor over a file with coupling.
Fig. 25 Hover over a file to see its connections to other files.¶
Interpreting change coupling depends on the context.
Widespread change coupling in a codebase can indicate tightly coupled logic and possible architectural issues, like in this example:
Making a change in one of these files often means making changes to
the others as well. It’s easy to imagine how these interdependencies complicate things
for developers who need to fix bugs or add features in any of those
files. In this case, change coupling is problematic.
Change coupling can also be positive, when it occurs between files that have a good reason for changing together. This is the case with test code, like in this example:
Fig. 27 Change coupling between a file and its tests.¶
There is coupling between SyntheticEvent.js and its test file:
SyntheticEvent-test.js. This is a positive sign that suggests that the tests
are being updated as the file evolves.
Hovering over files with change coupling causes the coupling lines to appear for as long as you hover.
This can be inconvenient if you want to zoom out or shift the viewport to see the rest of the map.
In that case, clicking will “lock” the change coupling visualization, leaving you free to continue to interact with the system map.
There are two other consequences when this happens:
When one file is “locked”, hovering over other files still works. In this case, the coupling lines are white, in order to differentiate them from the first set of couplings.
When a file with change coupling is selected, a third tab will appear in the sidebar, providing detailed information about the different change couplings.
Fig. 28 With a file locked, the change coupling sidebar appears¶
The “Degree” indicates the strength of the coupling. (For more information about how the couplings are calculated, see Change Coupling: Visualize Logical Dependencies. )
The “Coupling Source” identifies how CodeScene identified the change coupling. The system maps use two possible sources:
Commits: files are modified as part of the same commit;
Project management tool tickets: files are modified as part of the same ticket.
In general, commit-level change coupling indicates a stronger level of coupling.
However, changes made in separate repositories cannot be part of the same commit, so we then turn to ticket-level change coupling, which is not perhaps as “strong” but has the advantage of seeing couplings that would otherwise remain invisible.
For more information, see Change Coupling: Visualize Logical Dependencies. The change coupling
data used in the hotspot map is calculated the same way as for the
other change coupling graphs, with two exceptions:
Filters are not applied. This allows “positive” coupling, like for tests, to be visible.
The coupling threshold is fixed at 20% which will generally be lower than the automatic default in the change coupling graph.
Filter Hotspots interactively by Team or Developer¶
It’s common to have multiple development teams committing to the same codebase. The interactive hotspot map
lets you filter the view by team or authors. That way, you can limit the information to what’s relevant to your team:
Review your Findings: CodeScene’s Virtual Code Review¶
CodeScene’s virtual code review presents the code health findings, together with other module level analyses that let
you investigate any hotspot:
Fig. 30 View the Code Health trends for your hotspots.¶
The main advantage of using hotspots to guide improvements is that you’re able to
narrow down refactorings to a small part of the system. That in turn
will give you more time to tackle larger issues once you’ve made these
initial improvements.
To prioritize hotspots, CodeScene employs algorithms that
look at deeper change patterns in the analysis data.
CodeScene starts by identifying files with high change frequency and low Code Health.
Then CodeScene factors in other aspects.
Complicated code that changes often is more of a problem if:
The hotspot has to be changed together with several other modules.
The hotspot affects many different developers on different teams.
The hotspot is likely to be a coordination bottleneck for multiple developers.
This algorithm allows CodeScene to rank and prioritize the hotspots in your
codebase as illustrated in Fig. 31.
Fig. 31 CodeScene prioritizes the Hotspots in your code.¶
Red refactoring targets should be dealt with first. Remediating technical debt or code quality issues in
the red refactoring targets is likely to quickly produce a real return on your refactoring investment.
Use the hotspot’s Code Health to get a quick assessment of potential technical debt or maintenance problems as shown in
Fig. 32. We talk more about Code Health in the next section.
Fig. 32 CodeScene prioritizes the Hotspots in your code.¶
Once you’ve addressed those hotspots, the yellow hotspots become
interesting as well. A yellow hotspot is secondary target with lower priority than the red category.
By default, CodeScene calculates the code health of all the files in your code base.
Fig. 33 View the code health of each file in your codebase.¶
Viewing the full code health perspective also highlights the importance of priorities. Even if a module has a lower health,
that does not mean that it requires attention or immediate action. Code quality has to be put into perspective, which
is something CodeScene’s other analyses do. Let’s look at some examples.
First, use CodeScene’s priorities – our machine learning algorithms will sort out the relevant code health issues from
the ones with less priority:
CodeScene prioritizes the most relevant hotspots automatically.
Optionally, you can use the slider in the hotspot view to inspect the code health of the most active parts of the code:
Fig. 34 Filter the code health view based on development activity.¶
There’s a strong correlation between Hotspots and software defects.
When you come across hotspots with severe maintenance issues, there’s always going to be a trade-off: do we pay-off the
worst technical debt or should we continue to shoehorn yet another feature into the hotspot? Ideally, we would like to
know for sure that if we invest, say, two weeks into refactoring the code, then that effort will pay-off immediately. At the
time of writing, there’s unfortunately no way of looking into the future. What we can do instead is to look at the existing
costs and consequences of not doing any preventive and pro-active code improvements.
For this purpose, CodeScene comes with a Defect Density view. Since most organizations have a known and estimated number on
how much a defect costs, let’s use defects to predict the costs of any sub-optimal code we might find in our hotspots. Fig. 35
shows an example from CodeScene’s dashboard.
Fig. 35 CodeScene’s dashboard displays the bug density of the prioritized hotspots.¶
The statistics on the dashboard tells us the following things about the development costs in our codebase:
The prioritized hotspots only make up 1.2% of the total codebase, yet
we spend 12.5% of our development efforts in those hotspots, and
45% of all bugs that we detect and fix are in that small part of the code.
It can come as a surprise that almost have of all bugs, 45%, are
located in 1.2% of the codebase. However, this is an extremely common
situation and demonstrates the effectiveness of the hotspot as a
diagnostic took. This also has direct implications on the costs of
the whole system.
CodeScene’s Defect Density view shows how distributed our bug fixes are, which lets you correlate defects with hotspots as shown
in Fig. 36.
Fig. 36 Correlate prioritized hotspots with the distribution of defects in the codebase.¶
CodeScene needs a data source for its defect mining, and provides two different options depending on what data you have:
Use issue types to identify defects: This requires a that you integrate a PM tool with CodeScene. CodeScene will then
use all issues identified as defects for its statistics.
Specify this option in CodeScene’s Project management integration as described in Integrate Costs and Issues into CodeScene .
Use commit message patterns to estimate defects: If you have specific tags in your commit messages that can be used
to identify defects, then this is a good option. As a fallback, CodeScene can use a heuristic that you can override that with
more specific patterns for higher precision, as shown in Fig. 37.
Fig. 37 Configure a pattern to match defect information in your commit messages.¶
This configuration item can be found in the “Code Health” tab of the project configuration.
