Real-world software development methodologies?

Question

I'm currently learning for an exam in Software Engineering. However, I'm becoming frustrated by the apparently impractical and unrealistic techniques for producing high-quality code suggested in the lecture.
[edit:added this illustration from response to a comment]
For example, it was suggested that one could use linguistic analysis to design an application. That is, write down a use case in prose, underline all substantives and turn them into classes, underline all adjectives and turn them into attributes and so on. This is done to create an exhaustive UML diagram which can then be turned into code almost automatically. I would be pretty surprised if this would work in real life

There was however one chapter I enjoyed reading, which talked about what methodology Microsoft is using to develop their software. (Article: http://portal.acm.org/citation.cfm?id=255698, requires ACM membership)

So my questions is:
Are there other resources available which show how software development is done at large, successful companies? (successful meaning they deliver quality code in a timely fashion)

Also, are there any negative Examples available, e.g. examples of how software development is done in companies that regularly produce buggy and late releases?

Answer 1

Don't confuse a successful company with high quality product. And I'm not sure what 'impractical and unrealistic techniques' you are speaking of, but my guess is that they are being utilized to some degree or another by companies that do produce high quality code.

Answer 2

I thoroughly enjoyed the report Scrum and XP from the Trenches.

Answer 3

As an answer to your first question, I think that my absolute favorite pattern catalog: "Organizational Patterns of Agile Software Development" by James O. Coplien could be a valuable resource. This book is not about software patterns, is about people, a catalog for creating successful teams. A lot of stuff from the book is a case study of the successful Borland Quattro Pro for Windows development project.

"This is a remarkably wise book, full of pragmatic advice drawn from real projects. Ultimately, software development is a human experience, and Jim and Neil have captured the essence of that experience in this work." --Grady Booch

I didn't read it yet, but I guess "Dreaming in Code: Two Dozen Programmers, Three Years, 4,732 Bugs, and One Quest for Transcendent Software" by Scott Rosenberg should deserve to be read also.

Answer 4

Google for extreme programming, agile, waterfall and spiral model, RUP, RAD, software architecture, software development as a process, project management, software as design. You will also find many useful articles and books while doing so. Software engineering in real life is not exact, precise rocket science as taught in CS classes. It is a chaotic process with many possible approaches, that sometimes delivers. And there is no silver bullet.

(UML is a really nice way to visually represent some aspects of software design and the common way to start diving into something as useful as software engineering. But forget phrases like "which can then be turned into code almost automatically" which is marketing bullshit for tools or a placeholder in publications. Albeit you will hear it many-many times from now...)

Answer 5

We created OPEN/Metis a few years ago in the context of my company Neco, and using research outcomes from my previous work at the University of Santiago de Compostela in Spain. It was successfully used in over 20 projects, and still used today.

Answer 6

If you want to know how software is (or was) really produced at Microsoft, take a look at Show Stopper! which is about the original development of NT. Frightening stuff. And as a counterwieght to that, read Microsoft Secrets, which is all about their more recent development strategies - and after all, they are moderately successful at this.

Answer 7

Most software development projects use the (outdated but still widely used) waterfall methodology, i.e. a predictive approach, and many of them are failure (they are late and over budget) because the nature of software development isn't like mass manufacturing, things aren't really predictive.

Agile methodologies (which belong to the iterative and incremental approaches family) are an alternative to the traditional waterfall and are adaptive approaches, Scrum and XP being the most famous.

Answer 8

In addition to what has been already said, I would like to recommend the following books:

Code Complete 2 by Steve McConnell

The Pragmatic Programmer: From Journeyman to Master

Rapid Development: Taming Wild Software Schedules

97 Things Every Programmer Should Know

Coders at Work

Clean Code: A Handbook of Agile Software Craftsmanship

Though not all the books deal directly with your question, they all will help you in your journey of becoming a better software developer.

A great online resource that I would recommend is the Empirical Software Engineering and Measurement (ESM) Microsoft Research Group. The article titled Exploding Software-Engineering Myths is a good example of what you are trying to learn.

Answer 9

If you are skeptical and frustrated about how software engineering is taught, have a look at this paper:

• Twenty Dirty Tricks to Train Software Engineers

Many employers find that graduates and sandwich students come to them poorly prepared for the every day problems encountered at the workplace. Although many university students undertake team projects at their institutions, an education environment has limitations that prevent the participants experiencing the full range of problems encountered in the real world. [...]

Another must have reading, really insightful:

• Bits of evidence: What we actually know about sw. development and why we think it's true

I also enjoyed these two articles related to software engineering:

• Exploding Software-Engineering Myths
• Frequently Forgotten Fundamental Facts about Software Engineering

Academia realized that the gap between theory and practice. Software engineering is nowadays seen more and more as a human-centric activity rather than a technology-centric activity. Most lectures slowly start to reflect this reality.

It's however hard to teach the soft aspects of software engineering which relates to interaction between people, changing requirements, risk management, and external factors (financial crisis, product acquisition, etc.)

Software engineering is a broad topic. It's not only about software design and coding, but also about requirement management, software architecture, human computer interaction and usability, design principle and best practices, software configuration management, product management (or even product line management), quality insurance and testing, risk management.

It's now accepted that one of the biggest challenge in software engineering is to deal with unanticiapted changes. That is, we can not plan everything ahead -- goodbye waterfall! Hence the rise of agile approaches. There are many such processes and methodologies out there, including Scrum, XP, Test driven, Domain Driven, and RUP and its flavours.

I'm however doubtful how much success of a project can be related to the process itself. A good team produces good results no matter the process they use (and inversely). The question is then whether success is reproducible.

• No Matter What They Tell You, It's a People Problem
• Is Good Design Replicable?

I will conclude with UML. UML is great, but the usage I've seen in practice stops too frequently to class diagram. That's just a tiny subset of what can be modeled.

• We should teach modeling, not UML

I whish UML was removed from software engineering lectures and move into dedicated lectures, say, "software modeling". The focus of such lecture would be entirely on software architecture and modeling, while the software engineering lecture would be about team collaboration, quality and process.