Jun 01

Is .NET the holy grail of software development in the modern age!! (Part 1)

I have spent the last 20 years of my working life involved with software development in some form or another.  My first programming experiences were on home computers like the Sinclair ZX81, Spectrum and Commodore 64 where writing in 6502 or Z80 assembler using just a word processor, a cassette recorder and a dot matrix printer if you were lucky was the only tooling to hand.  The first program I ever wrote on the PC was written in x86 assembler the DOS ‘debug’ tool and was a simple terminal emulator for connecting to a VAX over DECNet!  I stepped up to the ‘C’ language soon after that and eventually to C++.  I have also developed a bunch of AVR, ARM and PIC microcontroller code for numerous projects too. Now days, pretty much any language is usable to some degree or another, and my software thinking in terms of achieving a specific task has become somewhat language agnostic.  As you can tell from my past though, I am someone with a slightly unhealthily liking for low-level programming and I have clearly had way to much spare time, late nights, coffee and pizza!

Anyway, back to the topic in hand. What does all that have to do with .NET you are probably thinking?  Good question – but before I answer that, I must clarify what I mean when I say .NET for the purpose of this blog post.  I am specifically talking about the lowest level of the .NET stack which is the CLR and the .NET framework base libraries.

My affinity with the inner workings of computers and my low-level programming experience has become a bit of a curse for me in the modern programming age.  When working with C and C++ this low-level understanding can be a real advantage but when it comes to the higher level languages like those provided in .NET one must to a large degree ignore how the machine works and think just about  the language and the idealism that is presented within its design.

I should just state that much of what I am about to say applies equally well to Java, PHP, JavaScript, VB and other interpreted language environments and frameworks as it does to .NET and of course, it goes without saying that this is all my own humble opinion so take it as you will. Disclaimer Ends— ;-)

.NET is an interpreted language. I say this because the ,NET code running is being managed by an execution engine which is best described as a byte-code interpreter – despite what the marketing hype may indicate with things like JIT, native code generation and so on, the very notion of “managed code” means this must be true. You are probably thinking, what is wrong with that? Well nothing in principle, but what is wrong though, is what people choose to use this type of software stack for.

There is no escaping that .NET is popular and has a great following – and by the way in my opinion it’s pretty good for some things too, I am not anti-.NET and I must say the development tools that Microsoft have created in Visual Studio are exceptional and by far my most favourite modern-day development environment (Apple, please take note :).  The problem is though, .NET is not ideal for everything – yet despite this, some people tend to hail NET as the next coming.  I can’t tell you how many organisations I have come across over the years who have set a “corporate policy” that states all applications must be based on .NET.  How can an organisation decide on such a policy, it’s crazy because it’s unenforceable if you happen to use Microsoft’s own products – many of which to this day are not developed in .NET (more on this later).  I suppose that is the power of the Microsoft marketing machine, best not sell this to developers said Microsoft, lets sell it to the executives instead.

Let’s take a look at what some of the key marketing messages put out at the time were in order to answer the “Why .NET” question: -

  • Managed code means fewer bugs,  reducing maintenance time by up to XX%
  • Mixed language environment allows all of your existing skills even across  multiple languages to work on the same project reducing time to market
  • Platform independent (really, that’s how .NET and the (then called) IL was touted almost 10 years ago, using .NET will ensure you have portable code!! Yeah right – where is that!)
  • Garbage collection, exception handling and other language features makes it easier to write code that works reliably.
  • The standard framework and foundation libraries mean your VB developers can transition to C# much faster.

The message here was not really aimed at programmers but at the management and executives of companies and development teams. Use .NET and you can standardise and save time, which will save lots of money and improve you competitive edge by reducing your time to market. .NET means your teams can write code in half the time and the software will work.

Clearly, this is the way to go right? Perhaps we should look at where .NET came from. It was not a brand new innovative idea that came to someone in a vision – it was simply a natural evolution of the old VB6 environment which was plagued with all sorts of problems that needed to be resolved in order to move forward.  Microsoft’s first stab at a component model was COM, which was an evolution of an even more ridiculous specification called OLE.  VB6 was used primarily within corporate application development world and things were messy, and for the most part COM was the problem but was also the keystone technology that held VB6 together.  Windows today is still plagued with the aftermath of COM nightmare that seemingly will never go away much to Microsoft’s wishes to the contrary I am sure.  VB6 supported components through COM but components were unreliable, difficult to work with and component vendors would write bad code that would bring the whole application down. Although for the most part, the third-party components were the problem, Microsoft did not help because COM was so badly implemented in the first place anyone writing a COM component would invariably get things wrong and the sheer complexity of the COM specification sealed its fate.

.NET with its managed code environment, assembly management and standard component library was clearly the headline wish list for VB7.  So really, .NET is an evolution of VB – a significant one granted, but an evolution none the less – this was more obvious in the early versions of the .NET framework back in 2002.  When VB6 was at its height, I remember its place was within the enterprise where corporate organizations developing their own internal applications, there were very few off-the-shelf products around that were developed using Visual Basic 6.

Reviewing the marketing messages above its pretty clear that Microsoft’s intention, initially at least, was to migrate their VB6 user community to .NET – their aim was the corporate in-house development teams, the ISV community also targeted only really adopted .NET later on as it started to improve.   The corporate in-house development market is a real sweet spot for .NET IMHO, and it’s probably the best environment for in-house corporate desktop and web application development teams, with the only real competition being Java.

Perhaps controversially though, I don’t believe that .NET is the right software stack for ISV’s to choose when developing the core of a professional, off the shelf and supportable software product – choosing .NET is far more advantageous to the ISV than it is to the customers that ends up using the product!

Why? Well, I have already exceeded my word count on this one so if I have not managed to send you to sleep already, I will explain what I mean in Part 2…