Ask Me Anything sessions: A unique opportunity to learn more about Sigrid. Sign up now!

System-level maintainability

The maintainability view is available for all systems by default. You can reach this view in different ways: Via the top menu, or clicking an a capability on the System or Portfolio Overview pages. See the system-level Overview page or portfolio-level Overview page.

The definition of what a system is, what it is comprised of, and how it is configured, are detailed in the pages on systems within Sigrid and the analysis scope configuration documentation page.

4 different views in the Maintainability tab

The maintainability section on the system level has 4 views:

  1. The Overview tab brings the main metrics together.
  2. The Component Dependencies tab visualizes architectural layering and connections.
  3. The Refactoring Candidates tab groups and prioritizes code that does not meet 4 star quality.
  4. The Delta quality view shows the impact of new code changes on the system for the selected period.

Maintainability overview

The overview page is shown below.

Below a detailed view of the metrics. A mouse-over on the “?” help icon explains what constitutes the overall maintainability rating. The maintainability score consists of several sub-metrics that range from 1 to 5 stars, with the range 0.5 to 5.5.Please be aware that the overall score is not an average of the submetrics. For a general introduction on these metrics, see the section on our approach.

For technical details on maintainability metrics, see Maintainability Evaluation Criteria on our website. Or a broader overview on our Reference page on our quality models.

Below the metrics overview, there is a shortcut to the Refactoring candidates link on this page. This can also be reached by the Maintainability tab.

Technical Monitor and Code Explorer

The Technical monitor button above the system rating brings you to an alternative (one might say, “legacy”) view of all the maintainability metrics and underlying source code. Its functionality and views will eventually be moved to Sigrid. Generally, its source code-level view is available in Sigrid in the Code Explorer.

Because the views, filters and sorting abilities between the “Technical monitor” and “Code Explorer” are different, please see a deserved elaboration on the Technical monitor page.

Component Dependencies

The Component Dependency view visualizes the dependencies between your application’s main components. The components follow from the system’s configuration.

Meaning of the dependencies

The arrows denote call direction within the code; a number on top of an arrow indicates the count of dependencies (that is, >1). Note that only calls will be shown that are identifiable as code dependencies (“static”). This excludes dependencies that may occur in production because of communication to frameworks or resources that are not explicitly defined in the source code.

Visualization options and filters

Different types of dependency antipatterns can be shown by toggling Visualize component entanglement.

Once activated, a legend will appear at the bottom describing the different types.

The legend’s colors denotes the severity of the antipattern:

For details on their specifics, see the Reference page on our quality models, specifically the Maintainability Guidance for Producers (on the SIG website).

If you click on an arrow in the graph, a page will appear where you can inspect the individual dependencies from- and to the selected components.

Note that the calls are shown for the direction of the arrow that you clicked at. If you want to inspect cyclic dependencies, also inspect the dependencies in the other direction.

A filename mouseover shows you the full path. Clicking on the file name will bring you to its source code.

Annotations

The annotation menu can present different data on top of the components.

This may include e.g. code volume in PM/PY (person-month or person-year equivalent).

Filtering

On the left-hand side column, you can filter dependencies per component and/or file level. This will show you a more detailed view of dependencies.

Refactoring Candidates

This view lists the top 100 findings per metric.

Clicking on a metric will expand the list, prioritized by the “severity” of the violation. This is a good approximation of technical risk. The order/prioritization of the findings cannot be changed, but their status can be. The default status is Raw. This is meant in the sense of “not yet curated by hand”. Setting another status may help you to filter findings. A finding can be set to Prioritize or Accept risk.

Setting a finding to Prioritize will show as Will fix

When you set a finding to Accept Risk, its status will change to Risk accepted and the finding will be hidden by default.

Findings with Risk accepted can still be viewed by using the filter. By default the filter is set to Will fix and Raw only.

The relevant filter is shown below.

Ordering of Refactoring Candidates

Refactoring candidates are sorted by risk impact. This is shown as maintainability risk categories, color coded as green-yellow-orange-red from lowest- to highest risk. Within each category, code is sorted by code volume (since volume is the common denominator for the maintainability metrics). See for more details the technical documentation.

As an example, the risk categories for Unit complexity as shown at the top of the page:

The exception in this ordering is Duplication, where no different degrees of risks are used for the rating calculation. They are ordered by duplicate size, where a duplicate may appear more than 3 times, in 1 or multiple files. This is visible next to the file names in the columns Same file and Same component.

The risk impact ordering is a good indication for prioritization of findings, but it may need a case-by-case analysis. Context is a defining factor, which is discussed below.

Dealing with Refactoring Candidates

Being refactoring candidates should be taken literally. It is not to say that every candidate needs to be resolved. No system is technically perfect (or it is not for long). Every metric has tolerances for violations of the risk categories, and these violations may be defendable.

The decision to refactor is essentially a cost-benefit trade-off. As a simplification this is determined by:

Questions to ask yourself dealing with refactoring candidates include:

Problem

Effort

Benefits

Sigrid as part of the Agile development process

For an elaboration of using and prioritizing maintainability findings within the development process, see the elaboration in the Agile development process document

Investigating system maintainability rating state and -changes

A typical approach and different options to investigate what is going on in terms of maintainability metrics are described below. Getting an initial overview is discussed here in most detail. The options to further analyze have their own respective pages and paragraphs and are referred in the text.

With the default treemap view, as an example, a large drop in Component independence may lead you to filter on change in that metric specifically over the chosen time period. With the following menu:

This results in the following overview colored by rating change impact on a green-to-red color scale:

To get an insight into the point of time of large changes, it may be useful to turn the default treemap into a trendline (change in the Chart drop-down menu) to see approximate when large changes have taken place. As an example, distinguishing between different metrics (System properties):

Assuming that this has been your first step into maintainability analysis, you can do several things next: