Ruminations on 20 Years of being a .Net Developer

.Net turned 20 years old and retrospectives are all over the web right now — so here’s mine!

First off, here’s what’s most important about my 20 years of being a .Net developer:

• I’ve made a good living as a .Net developer
• I’ve met a lot of great people over the years through the .Net community
• I’ve had the opportunity to travel quite a bit around North America and Europe to technical conferences because of my involvement with .Net
• Most of the significant technical achievements in my career involved .Net in some way (ironically though, the most successful projects from a business perspective were a VB6/Javascript web application and a Node.js/React.js application much more recently)

Moreover, I think .Net as a technology is in a good place today as a result of the .Net Core generation.

Those things above are far more important than anything else in this write up that might not come off as all that positive.

The Initial Excitement

I didn’t have a computer growing up, so the thought of being a software developer just wasn’t something I could have grasped at that time. I came out of college in the mid-90’s as a mechanical engineer — and to this day wish I’d paid attention at the time to how much more I enjoyed our handful of classes that involved programming than I did my other engineering coursework.

Fortunately, I came into an industry that was just barely scratching the surface of project automation, and there was opportunities everywhere to create little bits of homegrown coding solutions to automate some of our engineering work or just do a lot better job of tracking materials. Being self-taught and already working heavily with Windows and Microsoft tools, I naturally fell into coding with Office VBA, ASP “classic”, MS Access, and later to Visual Basic 6 (VB6) and Oracle PL/SQL. As the dot-com bubble was blooming, I was able to turn my “shadow IT” work into a real development role in a Microsoft-centric company that at that time was largely building software with the old Microsoft DNA stack.

I first worked with .Net to to an architectural spike with VB.Net in the fall of 2002 as a potential replacement to a poorly performing VB6 shipping system. That spike went absolutely nowhere, but I appreciated how the new .Net languages and platform seemed to remove the training wheels feel of classic VB6. Now we had full fledged inheritance, interfaces, and full fledged OOP without the friction of COM Hell just like the Java folks did! We even had real stack trace messages that gave you contextual information about runtime errors. Don’t laugh since that’s something most of you reading this take for granted today, but that was literally the thing that helped us convince management to let us start working with .Net early.

My second “real” job as a software developer was my introduction to Agile Software Development — and that’s when the initial excitement about .Net all wore off for me.

The early mainstream tools in .Net like WebForms were very poorly suited for Agile software processes like TDD and even CI (.Net was difficult to script on the build server back then). Many if not most .Net development shops had the software as construction metaphor philosophy where teams first designed a relational database structure, then built the middle and front end layers to reflect the database structure. The unfortunate result of that approach was to greatly couple application logic and functionality with the database infrastructure and further make Agile techniques harder to use.

So fine, let’s talk about ALT.Net for just a minute and quickly move on…

In the early to mid-2000s, and as many of us who were trying to adopt Agile development with .Net were becoming increasingly frustrated with the current state of .Net tooling, Ruby on Rails popped up. I know that Rails has probably lost a great deal of its shine in the past dozen years or so, but at the time it was brand new, Rails made ASP.Net development look like a giant turd.

Right at the time that the rest of the development world seemed to be changing and improving fast, the .Net world was insular and largely a monoculture and self-contained echo chamber revolving around Microsoft. The catalyzing event for me was the MVP summit in 2007 when the poor unsuspecting Entity Framework team did an introductory demo of early EF and were completely blindsided by the negative feedback that I and many of the original instigators of ALT.Net gave them at the time.

Long story short, we thought that the early version of Entity Framework was a complete disaster on technical grounds (and I still do to this day), and because of Microsoft’s complete dominance of mindshare in the .Net ecosystem, that we were all going to end up being forced to use it in our shops.

That experience led to initially the EF Vote of No Confidence (read it at your peril, I thought the wording was overwrought at the time and hasn’t aged well). More importantly, that was the beginning of the ALT.Net community that drastically changed my own personal path within .Net.

If you care, you can read my post The Very Last ALT.Net Retrospective I’ll Ever Write from a couple years ago.

I’m happy to say that I think that EF Core of today is perfectly fine if you need a heavy ORM, but the Entity Framework V1 at that time was a different beast altogether and best forgotten in my opinion.

ASP.Net MVC and .Net OSS

Scott Guthrie was at the first big ALT.Net open spaces event in Austin in the fall of ’07 to give the first public demo of what became ASP.Net MVC. More than ALT.Net itself, I think the advent of MVC did a tremendous amount of good to open up the .Net community to ideas from the outside and led to a lot of improvements. I thought then and now that early MVC was mediocre in terms of its technical approach, but hey, it got us off of WebForms and moving toward a better place.

I thought that OSS development really matured after MVC kind of broke the dam for alternative approaches in .Net. Nuget itself made a huge difference for OSS adoption in .Net — but, yes, I will point out that the community itself tried a couple times to create a packaging specification for .Net before Microsoft finally stepped in and sort of crushed one of those attempts by coopting the name.

Even outside of MVC I thought that .Net tools got quite a bit better for Agile development, and I myself was deeply immersed in a couple OSS efforts (that largely failed, but let’s just move on). .Net is probably never going to be a great for OSS development because of Microsoft’s complete dominance and their tendency to purposely or not, squash OSS alternatives within the .Net ecosystem.

Project K and .Net Core

I’m admittedly getting tired of writing this, so I’m gonna quickly say that I think that “Project K” and .Net Core maybe saved .Net from irrelevance. I think that Microsoft has done a tremendous job making .Net a modern platform. In particular, I think these things are a big deal:

• Being cross-platform and container friendly
• The dotnet CLI makes project scripting of .Net applications so, so much easier than it used to be
• The newer SDK project system made Nuget management so much better than it was before. Not to take anything away from Paket, but I think the newer project system plus the new dotnet CLI was a game changer
• The performance improvements that have been baked into the newer versions of the .Net framework
• The core, opinionated IHost, IConfiguration, IHostedService, etc. abstractions have done a lot of good to make it easier to spin up new systems with .Net. I think that’s also made it much easier for OSS authors and users to extend .Net

Granted, I routinely get irritated watching .Net MVP types gush about the .Net teams adding new features or ideas that were previously invented in other ecosystems or earlier .Net OSS projects that never took hold. So after a mostly positive post, I’ll leave off with a couple criticisms:

• .Net would be in a much better place if we weren’t throttled by how fast Microsoft teams can innovate if instead innovations from the greater .Net community had a chance to gain widespread adoption
• The MVP program is a net negative for .Net in my opinion. I think the MVP program generates way too much incentive to focus on tools from Microsoft to the exclusion of anything else. One of the drivers for founding ALT.Net to me back in the day was the tendency of .Net developers to jump on anything new coming from Microsoft without first considering if that new thing was actually any good.