7 replies
by realgem
Quality can be measured a couple of ways.
The first is how well the application will meet the requirements. This can be measured by determining how many individual requirements were not implemented (deliberately or otherwise). Some requirements are more important than others, so a weighting scale that is standard across all software projects should be used.
Another way is in terms of defects. A requirement may have been delivered in the software, but there may be a defect in the code, which keeps the requirement from being 100% effective. These defects could range from those that cause data-integrity errors to those that cause two fields to be misaligned. Defects should be rated on a severity scale. At minimum, you should rate them as cosmetic, serious, or critical.
Those are two ways that software quality is visible in the application. That is, visible to the end-user. Of course, there are also many ways to measure software quality that cannot be perceived by the user but are visible to another programmer.
Software quality also depends on how well the programming standards are followed, how well the code is documented, whether good programming practices have been used, etc.
In order to come up with a rating system, follow these steps:
1. Determine what software characteristics embody "quality". e.g. requirements delivered, defects, programming practices, etc.
2. Set scales for assessing how well the characteristics are met. e.g. defects are cosmetic, serious, critical. Requirements are critical, important, minor. etc.
3. Set weighted scores for each point in the scale.
4. Standardize the scores (e.g. use a percentage), perhaps in relation to the size of the application. This is important because larger applications will generally have more defects, and so a strict count will favour smaller apps.
by ramchandar1
by stan p
One way of looking at software quality, is to calculate the cost of the quality system.
If you analyse how much is spent on quality at different stages of the software lifecycle and express this as a percentage of the total cost, you will be able to get an idea of how much it is costing.
Such an analysis will show where the main quality costs are being incurred. This will then put you in a position to see where improvements could be made.
To carry out such an analysis you can use the Prevention-Appraisal-Failure model as a guide.
Prevention ? This covers the costs involved in the specification and design phase of the software.
Appraisal ? These are the activities involved during the development of the software. This will include, code inspections, peer reviews and software testing. This is the cost of checking the software before it is delivered to the customer.
Failure ? Once the software is delivered, any failures will generate costs for the supplying organisation. Such costs involve the diagnosing of the problem, bringing it back and analysing, re-working the software in development including code reviewing, re-testing and re-issuing. Such costs will be even higher if the fault turns out to originate at the design stage, prompting a more extensive re-work.
The three stages described above all carry costs for an organisation. However, each successive stage carries a greater cost for rectifying problems. Therefore an organisation which has a high cost of failure and a low cost of prevention, should consider investing in more specification and design work. This should improve the quality of the software produced. This improvement should be reflected at the appraisal stage, by the need for less code inspections and less internal re-testing of the software. This again should then be reflected in less software failures. The cost of software failure on a customer site is mu
by ramchandar1
by ramchandar1
by ramchandar1
This discussion is closed to new replies.
Trending IT Employment posts
Sort by Trending