Software innovation is often great news for business – yet delivering a new application which works for immediately is another matter entirely. New programmes can often take months of testing, with the potential to stretch costs further than anticipated.
New applications can unleash a wealth of new customer connections, improve a company’s online presence and automate tasks so hum-drum that human staff would rather drill their own fillings rather than carry them out. For the CIO, contractor hire and ROI must all be taken into account – not to mention potential barriers of legacy infrastructure and stubborn company culture.
The carefully crafted commands woven into the script will fail at the slightest logic error, mistype or incomplete path. As every developer knows, the computer is a stubbornly specific taskmaster and will not take kindly to requests made in the wrong form of language.
Best practise in the test driven development (TDD) philosophy is that tests must be short, sharp and to the point. The approach prioritizes unit testing (ie the testing of a class or a function) in isolation of the rest of the system.
On the flip side, some developers feel strongly that TDD is short-sighted, and design driven by unit tests produces a weak foundation for applications. A proportion of software engineers hold TDD to be as outdated as waterfall as an approach, with DevOps emerging as the champion for an increasing number.
Sound like a lot to bear in mind? Fear not, and bookmark this handy guide as your new go-to software testing guide.
Blackbox testing
This is the moment the developer has been waiting for: the preview of the complete application from the user’s perspective. Blackbox testing focuses on the front-end and is often used for software validation. As its name suggests, this approach puts the code script to one side and pretends it does not exist.
A difficulty here is that the developer knows what s/he wants the application to do so well, that the wood may not be visible for the trees. After further checks from “new eyes” on the project within the company, many firms opt for beta testing for a selection of their clientele, collecting feedback as a priority.
White box testing
In contrast to its dark companion, white box testing — also known as glass box or structural testing — makes use of the developer’s detailed knowledge of the application code’s logic. Pieces of the internal code are put to the test based on the coverage of code statements, conditions, paths, branches and so on. This is a crucial step because it highlights lines of code which are not being executed due to typos or faulty logic. Of course, this means the software tester must have an in-depth knowledge of the programming language employed.
White box testing is a broad category which can be sub-divided into two streams: unit testing and integration testing. It is predominantly used for validation testing.
Unit testing
Typically carried out by the programmer, unit testing runs individual software components to check how well each one has been constructed. This branch of experimentation can be further broken down into mutation testing, operation testing and execution testing. This family of approaches could necessitate producing test harnesses or test driver modules.
Execution testing
This white box method has three sub-types: statement coverage, branch coverage and path coverage. Statement coverage is possibly the most granular type of software test: the validation of whether a line of code will execute. Following on from this is branch coverage, also known as decision coverage, where the “true” and “false” lines of IF statements are followed through for validation. Finally, path coverage comprehensively tests all paths of the programme and adds in necessary statements if a condition should fail.
Integration testing
Particularly relevant to client/server and distributed systems, integration testing runs all the integrated modules to verify their combined functionality after integration. There are three major kinds of integration testing: top-down approach, bottom-up approach and the hybrid approach. For instance, incremental integration testing is a bottom-up method, employing continuous testing when a new capability is added to an executable.
The evolving CIO and the finance frontier
Enterprise innovation KILLED as developers face monotony
Dynatrace ‘Untouchables’ zone in on AI, analytics and digital experience
Performance/stress testing
A clever method of putting the code through its paces is by lobbing as much data at the programme as possible and seeing when it cracks under pressure. Various load tools are utilised to check whether the system meets expected performance standards.
Much the same as hardware testing, the script writer or software expert will overload the application to see when and why it breaks. Examples of this mischievous endeavour include overly complicated database queries, over-amped data input or simply breaching specified capacity.
Vulnerability testing
In a world of IoT device proliferation, smart technology and growing demand for AI solutions, checking a new application for cybersecurity weaknesses is now potentially more critical than ever before. With malware, phishing attacks, and even chip vulnerabilities on the cards, extensive testing should seek out security flaws like the company’s IP depends on it. As, in some cases, it just might turn out to.