1. Technical Debt

Concept

Technical Debt is used to address the non-quality of a software project. It is evaluated in time unit (man day), or actual cost (monetary value), based on the remediation cost of all default which have been found in the project.

Technical Debt represents the effort to fix all quality issues in the project.

Utility

Using Technical Debt helps anticipate future issues the development team will face. The more debt a project has, the more difficult the development will be. High debt often reveals a low ratio of “new features”, as team spends its time in understanding and fixing current issues.

SWAN tech debt
technical debt versus features

Technical Debt = Maintainability?

No, Technical Debt concept is not limited to Maintainability, and can be extended to other ISO quality characteristics: maintainability, portability, reliability, security, efficiency.

However, quality issues other than Maintainability are most of the time addressed during development because they are part of the delivery requirements (ie: no reliability issue shall be detected in the delivered product…).

Maintainability is often considered as rather “Nice to Have” than “Mandatory”. As a consequence, technical debt is (wrongly) associated to Maintainability.

Computation

Technical Debt computation is based on quality issues provided by Static Code Analysis. Each issue is associated to a remediation cost, mapped to a time effort.

SWAN tech debt computation
Technical Debt computation

Formula:

sum_(All__issues)=(#violations * remediation cost)

Remediation Cost:

Remediation Cost

Minutes

Low

10

Medium

30

High

60

Huge

480

Squore uses the ISO mapping to provide Technical Debt trend according quality breakdown.

SWAN tech debt trend
Technical Debt trend according to ISO characteristics

Technical Debt is based on the findings provided by the static analyzers. Its value may vary depending on the analyzer which is used (in case they do not generate the same amount of issues). Activating a maximum number of programming rules during static code analysis increases the confidence level in the technical debt results.

How to address Technical Debt with Squore

Find the right project

At project portfolio level, Squore provides a Technical Debt chart highlighting the distribution of all projects according their technical debt density (=average of technical debt per 1000 lines of code). Interpretation: • Projects in the top area have a higher Technical Debt Density meaning they are more difficult to maintain, • Projects in the right area are bigger (wrt. Line counting), • The bubble size is proportional to the "number of programing rules which have been activated during static analysis". A larger bubble means a higher confidence in technical debt value.

SWAN tech debt quadrant model
Technical Debt at portfolio level

Find the right function

Inside a project, Squore provides a Technical Debt chart highlighting the distribution of all modules (i.e functions) according their technical debt and their violations density (density of issues weighted by severity). Interpretation: • Modules in the top area have a higher Technical Debt meaning they are more difficult to maintain, • Modules in the right area are bigger (wrt. Line counting), • The bubble color shows the severity of issues inside the components.

SWAN tech debt quadrant module
Technical Debt - Module distribution

2. Test Strategy

Concept

The Test Strategy intends to help development team increase the reliability confidence of their software project. It is more and more difficult carry out exhaustive testing activities, as the size and complexity of the code grow. With the Test Strategy you can:

  1. Reduce the scope of the code which needs to be tested

  2. Define code coverage expectations for those component which need to be tested

As a result, Squore provides the Code Coverage Compliance KPI which represents the ratio of components which comply with this Test Strategy.

Settings

To Be Tested

Squore applies an algorithm to include/exclude modules which should be integrated in the Test Strategy. Parameters are:

  • Cyclomatic Complexity (VG)

  • Nesting Level (LEVL)

  • Number of non-cyclic paths (NPAT)

  • Vocabulary Frequency (VOCF)

  • Code Stability Index (SI)

Using these parameters, the most complex and unstable code are identified. Increasing the test on this code will increase reliability and reduce the risk of delivery. Consequences:

  • Modules which are not part of the “to be tested” list will be “ignored” in the Code Coverage Compliance KPI

  • Modules which shall be tested will be evaluated according their coverage (Statement, Branch and MCDC) with regards to their safety level (ASIL, SIL …)

Coverage Thresholds

The code coverage thresholds can be tuned according the type of coverage and the safety level. Here are the default settings:

SWAN test strategy cov settings
Coverage thresholds

How to address the Test Strategy with Squore

How to setup the parameters

Parameters are available at project creation time. The “To be tested” list parameters are available in the “Test Strategy” section.

SWAN test strategy settings
Test Strategy - settings

Notes: • Each of the ‘TO BE TESTED’ parameters can be disabled (by setting the value ‘-1’) • It is possible to disable the impact of test strategy on the Code Coverage Compliance KPI. “Code Coverage thresholds” are available in the “Test Coverage Thresholds” section.

SWAN test strategy coverage form
Test Strategy - Coverage Thresholds

Note: it is possible to disable a dedicated type of coverage (Statement and/or Branch and/or MCDC). For instance if the team just wants to assess the statement coverage only, Branch and MCDC can be disabled.

How to setup the Safety Level (=Critical Factor)

Squore allows defining the critical factor for every component in the project.

Using the GUI:

  • Select an artifact

  • Open the Form tab

  • Define the critical factor

The critical factor will spread to all “children artifacts” (i.e. all modules will automatically inherit the value of the file). In addition, it is possible to overload an inherited value.

SWAN test strategy asil def
Test Strategy - Safety Level definition

Using a Text/Csv file import:

It is possible to provide a csv file during the project creation which contains the Critical Factor information. Create a csv file as follow (1=level A, 2=level B, 3=Level C, 4=level D):

SWAN test strategy asil file
Test Strategy - sample of "safety level" csv file

Feed this file to the “csv tag import” data provider:

SWAN test strategy asil dp
Test Strategy - Safety Level Data Provider

The Code Coverage Compliance Treemap

The treemap highlights components according to: • Their code size (=size of the treemap zone) • Their code coverage compliance (=color of the treemap zone)

SWAN test strategy treemap
Test Strategy - Coverage Compliance Treemap

Interpretation: • Grey zone means the module is excluded from the Test Strategy • The color code indicates how well the component is tested regarding to expectations defined in the “Test Coverage Thresholds” form

SWAN test strategy cov scale
Test Strategy - Coverage Compliance scale

Note: Safety level is implicitly taken into account in this evaluation. Whatever the safety level is, Squore highlights the distance to the coverage objectives.

3. Rule Compliance

Concept

Rule Compliance indicates how well the project development team follows the coding guidelines. A rule is not compliant if at least one unjustified violation is detected. If a violation is relaxed/justified using the “relaxation feature”, it will not impact rule compliance. For this KPI, Squore only takes into account rules which nature is categorized as “Non Conformity”.

SWAN rule compliance findings
Rule Compliance - Findings

Formula

Rule Compliance = (#violated rules / #Verified Rules)

Where: • “Violated rules” is the number of rules for which there is at least 1 violation • “Verified Rules” is the number of rules which were checked during static code analysis

Side effect of external data provider

The list of “verified rules” depends on the static code analysis and thus, depends on the external tools results which are imported in the Squore project with Data providers.

Example:

  1. Using ‘Squore Analyzer’ Data provider only: Total verified rules is 19 (for C language)

  2. Using ‘Squore Analyzer’ + ‘PCLint’ Data providers: Total verified rules is 152.

Rule Compliance may drastically vary, depending on the static code analyzers used on the project. Note: Squore dynamically create the ruleset based on the selected data providers. For instance, MISRA Checker will not be counted in the “verified rules” if no MISRA analyzer are called.

SWAN rule compliance rating impact
Rule Compliance - Rating Impact of Data Providers

The reason comes from the “verified rules” which drastically changed:

SWAN rule compliance rating before
Rule Compliance - Rating with Squore only
SWAN rule compliance rating after
Rule Compliance - Rating with Squore and PCLint

Settings (Ruleset template)

After selection of the data providers, Squore provides an interface to tune the ruleset in order to enable/disable specific rules.

SWAN rule compliance template ruleset
Rule Compliance - Ruleset Template

It is possible to create a template to share this ruleset configuration between projects.

Relaxing a violation

Relaxation inside Squore GUI

It is possible to relax a violation or a group of violations within Squore.

SWAN rule compliance relaxation
Rule Compliance - Findings Relaxation

*Note: Make sure you have selected the “Current” version of the project. Indeed, it is not possible to modify/update a baselined squore version. *

Relaxation imported from external tools

Squore can manage justifications which have been provided outside Squore environment. This mechanism is handled by the data provider, which detects justifications from the data it analyzes, and injects them into Squore.

4. Violations Density

Concept

The Violations Density intends to highlight the part of the code with the most violated rules (of “Non Conformity” or “Risky Construction” nature). The density takes into account the severity of the issues which are detected: the density is weighted by the findings severity. This allows to highlights either:

  • Code with blocker/critical issues

  • Code with a lot of minor/major issues

In both cases, remediation shall be performed to improve code quality.

Formula

Violation Density = (sum(issue * severity)) / code size)

Where: • Code size is expressed in KEloc (Effective Lines /1000) • Severity weights are:

Severity

Weight

Minor

0.1

Major

5

Critical

20

Blocker

100

The rating is based on the following scale:

SWAN violation density scale
Violations Density - Scale

A component which is rated “G” has in average the equivalent of 10 blocker issues every 1000 lines.

The Violation Density Treemap

The treemap highlights the component according to: • Their code size (=size of the treemap zone) • Their violation density (=color of the treemap zone)

SWAN violation density treemap
Violations Density - Treemap

5. Cloning and Duplication

Concept

Cloning and Duplication KPI intends to focus on the code that should be reworked. It basically addresses architecture design and implementation. The KPI is computed thanks to Squore Analyzer metrics. These metrics highlight duplicated block and similar algorithms.

Algorithmic Cloning Vs Duplication

Duplication is directly computed from the source code textual analysis. Cloning on the other hand is related to the source code algorithm analysis.

The analyzer can generate both metrics and findings related to cloning and duplication. The quality model embeds 2 indicators: Inner Cloning and Outer Cloning.

  • Inner (or Internal) Cloning is related to cloning within the artifact ⇒ is the artifact well designed?

  • Outer (or External) Cloning is related to cloning outside the artifact ⇒ is the artifact cloned with another artefact?

SWAN cloning concept
Cloning - Overview

Metrics

The analyzer produces the following metrics:

  • ICC = Inner Cloned Code = Number of duplicated line of code within the artifact

  • CC = Cloned Code = Number of duplicated line of code outside the artifact

From these information the Inner and Outer cloning ratio can be computed

ICCR = 100 * (ICC/LC)

CCR = 100 * (CC/LC)

Findings

In order to help find cloning and or duplication instances, Squore generates findings that are reachable from the findings tab.

  • R_NOCFTC - No Algorithmic Cloning

  • R_NOCC - No Code Cloning

  • R_NORS - No Repeated Substrings (Block duplication)

  • R_NOCAC - Consider refactorization (Artifacts contains too many clones)

Example:

SWAN cloning findings
Cloning - Findings

6. Test Gap Analysis

Concept

The Test Gap Analysis highlights the GAP between the code changes and the Test results. Squore combines Test Results and stability of the code (SI) to classify modules in different categories:

SWAN test gap legend
Test Gap Analysis - Legend

How to address the test gap with Squore

Squore classifies modules according different criteria: Code is stable (based on source code changes: added/removed/modifed lines) Associated Test is Passed Code coverage complies with the test strategy Associated Test is up to date

As a result, Squore provides a Test Gap Treemap.

SWAN test gap treemap
Test Gap Analysis - Treemap

7. Self Descriptivness compliance

Concept

The “Self Descriptivness” KPI intends to highlight components in the code which are not well documented. The KPI analyzes the comment on different criteria:

  • Comment Quantity (evaluate the comment size regarding the module complexity)

  • Comment Quality (detection of “commented-out” source code)

  • Comment Style (check for documentation programming rules violations)

  • Comment Header (look for header comment)

Settings

During the project creation, it is possible to enable/disable the criteria in the “self descriptivness” section.

SWAN sdescr settings
Self Descriptivness - settings

How to track comment compliance with Squore

Squore provides a dedicated “highlight” which lists the modules and their compliance regarding self descriptiveness criteria.

SWAN sdescr highlights
Self Descriptivness - Highlights

8. Complexity

Concept

The Complexity KPI intends to assess the project’s risky components regarding their complexity metrics. The indicator takes into account several criteria.

SWAN complexity criteria
Complexity - Criteria

The Complexity indicator aggregates these metrics.

A Class or Function is considered as complex if at least half of these metrics do not respect the expected threshold. At project level, Squore provides a Complexity indicator based on the Volume and Distribution of the overall complexity.

SWAN complexity rating
Complexity - Rating

Formula

Distribution = (#Number of Complex "Function|Classes" / #Total "Function|Classes")

Volume = (#Size of Complex Function / #Size all Functions)

How to track the complexity with Squore

Squore provide a treemap which distribute the modules according to their complexity.

SWAN complexity treemap
Complexity - Treemap

Squore also provides “complexity highlights” which sort modules by complexity.

  • List of all modules, sorted by complexity:

SWAN complexity highlights all modules
Complexity - Highlights: all modules
  • List of all "unstable" modules, sorted by complexity.

SWAN complexity highlights unstable modules
Complexity - Highlights: "unstable" modules

Squore combines the complexity and the Stability of the component over the Monitoring Period.

9. Monitoring Period

Concept

The Monitoring Period defines the timeframe which is used to evaluate stability of components. It is defined in “number of days” or “number of Squore versions”.

Squore analyses the history of component Stability (SI) and determines if the component has changed during this monitoring period. This is very useful in different cases:

  • Test Strategy: in order to detect if how long components have been stable

  • Unstable Complexity highlight (in order to list only components which have changed in the monitoring period)

  • Unstable Code Coverage highlight (in order to only list components which have changed during the monitoring period)

  • Test Gap Analysis (in order to detect which components are unstable)

Settings

During the project creation, , it is possible to change the Monitoring Period parameters:

SWAN monitoring period settings
Monitoring Period - settings
  • Define the monitoring period unit The time period can be defined as a number of days meaning the “n” days before the last Squore analysis or as a number of “n” squore versions regarding the last Squore analysis. By default, the “number of days” unit is used.

Choice of the unit may depend on the development process maturity and the level of automation of the Squore analysis (ie, daily analysis in continuous integration vs. manual trigger analysis from the GUI).

  • Define the monitoring period The monitoring period duration can be set, depending on the monitoring period unit