As
most of the software developers know, software engineering is not just
"coding". It's a complex process which requires engineering vision,
analytical thinking, designing and software process management supported
with tools. This article contains some important approaches, techniques
and tools to improve software production quality.
Software quality depends on time and cost of course. But
you can be sure that spent money and time will return as so much more.
Quality standards will also be used for upcoming projects and increase
company growth speed on midterm.
- Choose a suitable process model and apply it correctly
As you know there are software process models like agile, scrum, iterative and incremental etc.
Choose one of these according to the project type. You don't have to
perform all rules strictly, you can modify most of them. The point is
"using a process" here, for a systematic and ordered development.
- Control the version of each required source
Using
a version controlling system is a must. Even if only one developer
exist, it must be used. Historical data, versioning, merging, ... is
very important for increasing productivity. Otherwise developers will
wait each other, can not detect performers and lines of previous
changes, can not perform versioning systematically etc. SVN, CVS and TFS are used for this purpose.
- Track issues with easy-to-use tools
Parallel with using a process model, issue tracking tool usage is essential. JIRA-like
tools are useful and easy-to-use. By tracking issues, you can track
productivity and software growth and get production reports easily.
Besides, those tools can be used to monitor developer work-hour
productivity.
- Perform and manage documentation
Perform
documentation as required (e.g. %10 of total production time). No
documentation is never a good solution as much as excessive
documentation. It may include code documentation, requirement
specifications, design documents, test documents, user manuals etc.
Those documents are needed to be managed and shared also, by using
version controlling tools or web based platforms (like Confluence).
- Use dependency management tools
Managing
dependency libraries (library projects, jars, DLLs etc.) are a big
problem especially for big projects. Configuring libraries to run the
application after each release or after each project check-out is a
hell. Use a dependency managing tool like Maven or at least auto build/copy script tools like Ant.
- Use continuous integration
Building,
deploying and versioning software is a big problem. Its time consuming
and reduces productivity. Because of these, use a continuous integration
tool (Hudson for example) and integrate it with dependency management or build tools (like Maven, Ant,
..). The tool may be configured to perform a build on each commit, on
clicking a button manually or on predefined periodic times...
- Perform testing and integration testing constantly
Testing
is very important for software quality. Test documentation which may
consist test scenarios, results and relations with issues is required.
Also, testing (UI testing, integration testing, ...) must be performed
constantly and periodically. Even if changing a single line of code may
crush the whole system or crash a hidden functionality. for example, JUnit is very popular for Java applications.
- Perform unit testing and automatize it
Unit
testing is as important as the other testing methods. Unit tests
provide pre-detection of most of the problems. By performing qualitative
unit testing, time consumption for other testing methods also
descreases. Automatizing these tests using continuous integration tools
or at least command line tools (Hudson with Maven for example) are important to keep software consistency and reliability.
- Collect metrics from production and use results
Coding
metrics (e.g. line of code, abstraction ratio, cyclomatic complexity,
...) gives us some good viewpoints about software. For example, by using
line of code maybe we can't (or we mustn't) determine the productivity
of a developer; but we can determine the growth speed of software
monthly. Complexity-like metrics tells us design errors before
deployment. These metrics can be collected by tools or plug-ins (e.g. CodePro Eclipse plug-in).
- Follow best practices of coding and control with tools
There is no "golden rule" suitable with all software projects, but there are best practices for project management, architecture, designing, coding, testing for most situations.
Performing those rules will increase the quality. For example, you can
define rules for code production (about indentation, commenting, magic
numbering, paranthesis etc.) and monitor convenience automatically and
periodically with external tools or plug-ins like Maven Surefire Report plug-in.
Software Engineering
