Monday, 2 July 2018

Reflections on Alt.NET Birmingham 2018


Around ten years ago, a bunch of developers got together in London under the banner of ‘alt dot net’ to talk about code, the universe and everything. Ten years doesn’t sound like much, but it’s actually quite amazing to look back on how much has changed. Back at that first alt.NET unconference, a handful of people in the room had a first-generation iPhone. I don’t think anybody had an Android device yet. Stack Overflow didn’t exist. Most of us weren’t on Twitter yet — In fact, my first-ever blog post was a writeup of that first alt.NET UK unconference and I didn’t join Twitter until three months later

Most .NET developers in 2008 were still running Windows XP, apart from the unlucky few who’d upgraded to Vista and couldn’t work how to go back. Visual Studio 2008 was the new shiny. Lots of shops were still hosting on Windows 2003 or even Windows 2000 Server. That first wave of alt.NET was a bunch of developers who found C# and Visual Studio to be a powerful and productive platform for building software, but a platform that came bundled with a particular set of ‘best practises’ that weren’t necessarily the way we wanted to be working. The .NET community, such as it was, was heavily concentrated around Microsoft’s own conference and events, and there was very little discussion in that community around ideas like extreme programming, agile, model-view-controller unit testing, browser automation, dependency injection… you know. Ideas that are widely acknowledged here in 2018 to be, if not magic bullets, then well worth knowing and understanding how they might apply to the work you’re doing.

And now here we are, ten years later, on a glorious sunny day in Birmingham. Twenty-odd developers getting together on a Saturday under that same Alt.NET banner to talk about… well, whatever we want to talk about. That’s the beauty of the unconference format. The attendees create the schedule on the day. There’s no prepared presentations, no speakers, no keynotes or programme committee — you come along, you bring your questions and ideas and the things you want to talk about, and we see what happens.

An hour of spirited discussion and a LOT of post-it notes later, and we had an agenda — and what a wonderful range of topics:

Messaging and reference data, Blazor, XAML, WebAPI and EF Core and OData, liberating structures and agile processes, quantum computing and Q#, the SAFE stack, devops for mobile apps, Docker and Kubernetes and network security, gRPC and Service Mesh, dot net command line tooling, “are micro services dead?”, functional programming (and functional-style programming in C#), running dot net core on Linux, Systems Thinking, building JavaScript services in .NET Core, diversity and inclusivity, risk based software architecture, Xamarin and F#, and a crash course in Git and GitHub.

One of the great things about the unconference format is that you’ll end up with a bunch of curious people in a room wanting to learn about something… but ‘cos there’s no speakers and no ‘experts’, we’ll just fire up a laptop, find a quickstart or a ‘hello world’ tutorial or something, and start hacking on it. So we started off the day with a mob programming session on Blazor — one of the gang had a half-finished experimental chatbot they’d been working on, so we got that up on a big screen and played around with it for an hour until we had asynchronous callbacks and model binding working.

The next session I went to was one I’d proposed about quantum computing — I saw an intro session from Anita Ramanan and Frances Tibble at DDD13 in Reading last week, and I’ve been itching to download Q# and try it out… and so, in the absence of any quantum computing experts to tell us the answers, we had another group hack session with a laptop plugged into a big screen. What I found really remarkable about this session was how easily we got the whole thing up and running. I’m running the .NET SDK on macOS, using VS Code and JetBrains Rider, and it only took a few minutes to download the quantum computing preview, install it, fire up a ‘hello world’ app (or the Q# equivalent thereof) and start playing around with simulating quantum entanglement.

Next up was a freeform discussion all about command line tooling — gulp, grunt, yeoman, yarn, bower, npm, the dot net cli templating system — how did we get here, what did we like, and why the hell does running ‘dotnet new mvc’ drop a bunch of .something.bower files into your repository? We talked a lot about boilerplate code — about the days when File > New > Project would drop a few hundred lines of .designer.cs and .csproj files into your solution, which was never intended for human consumption, but you could guarantee that eventually SOMETHING would go wrong with one of those files and you’d have to roll your sleeves up and work out how to fix it. And we talked about how for people that have experienced this, Microsoft’s move towards a far more minimalist project file syntax is a Good Thing — but that alongside that, there’s a whole bunch of complexity that we’ve adopted wholesale from the npm ecosystem for managing web assets and front-end dependencies. We talked a bit about Andrey Taritsyn’s BundleTransformer, and the idea of treating SASS, SCSS, TypeScript et al as part of your solution and relying on the .NET pipeline to compile them at request time, and how the npm/node mindset has encouraged developers to regard these sorts of concerns as compile-time rather than runtime operations. The consensus — amongst our small and opinionated but by no means authoritative group — was that there’s still a pretty steep learning curve for the old-school .NET developer trying to understand modern web build tooling, but that it’s generally all getting a lot better.

I also made an honourable mention of the latest .NET CLI tooling by describing how an hour previously I’d installed the Quantum Computing SDK on macOS and got a couple of demos up and running, without having to reboot into Windows or fire up Visual Studio, and how that was actually quite awesome.

We broke for lunch — which was entirely thanks to Ian Russell, who I gather ended up not only organizing but also subsidising quite a lot of the event. For which we are all eternally grateful, but if there’s anyone out there whose company might be interested in supporting this kind of event, here’s a bit of advice for you. Community groups and events like this can make a few hundred pounds go a long, long way, but if your corporate sponsorship process means filling out forms, signing waivers and NDAs, submitting corporate investment opportunity proposals… the sort of people who run events like this are unlikely to appreciate your ‘support’. Work out how to support this kind of event with £500 out of petty cash, or offer to pick up the bill for lunch or something, and we’ll just say nice things about you and tell everyone on Twitter how grateful we are. You can get a lot of good will and positive buzz (not to mention sandwiches!) for less than it costs to hire a .NET contractor for one day — but when it comes to this kind of thing, please remember that your accounting procedures are your problem and not anybody else’s.

But I digress. Lunch was delicious and Ian, next time you’re down our way, the London .NET gang are taking you out for dinner to say thank you.

And on to the afternoon. First up was a session about functional programming — what is it, why does it matter, and how do you do it in C#? Partly inspired by some of Mike Hadlow’s blog posts about refactoring object-oriented C# into a more functional style, it was a great discussion; with some folks in the room who have very much embraced the functional paradigm, some folks who have spent years dabbling in F# but still prefer C# for everyday development, and some folks who haven’t really encountered functional programming but have heard a lot about it and would love to know why it sounds so important. Ian Cooper shared the ‘devil’s argument against F#’ — “I’m not sure I agree with all of this, but I know it well and I think it’s useful to share it” — about how pure functional programming is never going to cross the chasm from the bleeding edge to mainstream adoption; about how a lot of the ‘textbook’ benefits of functional programming are that it scales easily across multiple cores but that as an industry we’ve very much embraced the idea of scaling using services and containers rather than building monolithic applications that scale effectively on high-performance hardware. I pointed out that this notion of ‘scaling’ is very much rooted in ideas around scaling straightforward CRUD processes to cope with massive volumes of users — the predominant computational model of high-traffic web applications — and that there’s still lots of challenges around simulation, video rendering, gaming and machine learning where scaling a local process across multiple cores is still the most compelling route to improved performance. We also briefly discussed the dynamic that exists between the F# community and the rest of the .NET ecosystem (TL;DR: it’s mostly very friendly and lovely, but there’s still a perception that the functional community can be a bit elitist when it comes to sharing ideas with their object-oriented cousins).

There were also several recommendations to check out Scott Wlaschin’s book ‘Domain Modelling Made Functional’, both for an excellent overview of domain-driven design covered in the first few chapters, and for some worthwhile insight into the value of using functional programming patterns in your domain models.

Next was a session that ended up on the board as a combination of ‘diversity’ and ‘attracting new speakers’ — but ended up running a broad gamut from conferences and community, to speaker tips, suggestions and ideas for encouraging more people to get involved in what we do. I’m the first to admit that .NET has a massive problem with diversity — the vast majority of .NET developers in the UK are middle-aged white guys, and whilst there’s many brilliant speakers, engineers and evangelists on our platform who don’t conform to that particular stereotype, if you turn up to a .NET user group pretty much anywhere in the UK, it’s going to be overwhelmingly white males of a certain age. We talked about this at some length, including the observation that a lot of the people who are still passionate about .NET as a platform are people who have been here since the beginning, since the days of Visual Studio .NET and C# 1.0, and there’s not a whole lot of people who have made a conscious decision to embrace .NET as a development platform who didn’t already have some kind of investment in the Microsoft/Windows/SQL Server platform. The session convened around a question — how do we get more people looking at .NET as a development platform? It’s free. It’s open source. It’s cross-platform. It runs on macOS, Linux and Windows, it’s got excellent hosting and infrastructure support from multiple cloud providers… so why aren’t we seeing more interest in it? (When we figure out the answer, I’ll let you know. Promise!)

Finally, we wrapped up with a park bench discussion about “The .NET Renaissance — Where Are We?”. Following on from a series of blog posts and conference talks (and at least one DotNetRocks podcast) that various members of the .NET community have shared over the last eighteen months or so, and overlapping a lot with some of the questions and ideas from the session on diversity and inclusivity, we talked about .NET — past, present and future. Ian Cooper shared some slides, graphs and stats based on various metrics — recruitment websites, TIOBE — along with a healthy disclaimer that all programming language metrics were probably wrong and there’s “lies, damned lies, and percentages”. We talked about how Unity is creating a path into programming for a lot of first-time developers who are interesting in creating games and immersive 3D environments, and how the increasing interest in devices like GearVR and Hololens is only going to encourage this. Jim Bennett (who I should point out was here sharing his own opinions and not any official Microsoft position) spoke about how the diversity of platforms and runtimes almost guarantees long-term investment in C# as compared to platforms like Flutter/Dart, which are still tightly coupled to specific hardware and OS platforms. .NET has got some absolutely first-class tooling support. It runs on a huge number of devices — not just Windows, Linux & macOS, but thanks to Xamarin it’s now running on Android, iOS, Samsung’s Tizen platform (yeah, did you know your Samsung smart TV can run .NET Core applications?) — and with companies like AirBnB talking openly about the challenges they’ve had trying to use frameworks like React Native to deliver cross-platform solutions across web and mobile, it’s clear that there’s still a big unsolved problem around building cross-platform consumer apps, and that .NET Core and Xamarin Forms is working really hard to offer a compelling solution to that problem that could turn out to be the incentive that attracts a new generation of developers and startups onto the platform.

We took a few moments to plug a couple of other .NET events that are coming up in the UK over the next few months — the ProgNET conference and tutorials that we’re running at Skills Matter in September, and DDD East Anglia, which is another free community event that’s taking place in Cambridge in September.

And then, after eight hours of inspiring conversations and insightful observations, we went to the pub to reflect on how the day had gone. One thing that stood out was that, asking people how they’d heard about the event, there was very little correlation. Many of the people who came along today heard about it through word of mouth, and despite promoting the event widely on Twitter, email and social media, we probably knew a few dozen people between us who would have come along if they’d known it was happening. Asking people to give up a Saturday to come along to an unstructured conference is a bit of a big ask, sure — but actually, I think it’s one of the things that makes the unconference format work well. If you ran an event like that on a Friday, you’d probably get a lot more signups from people who persuaded their boss to give them a day off to go to a free conference… but you’d probably also get a lot more people turning up late, sitting quietly without really contributing anything, and I daresay sloping off the pub at lunchtime and not coming back. Most of the people who came along today weren’t based in Birmingham, and even though the event itself was free, somebody who gives up a day of their time and pays out of their own pocket to travel to an event like Alt.NET is exactly the kind of person you want involved.

One thing was overwhelmingly evident, though. If you’d taken today’s agenda back in time to 2008 and showed it to the people who came along to that first Alt.NET UK event, they wouldn’t have understood half of it (“coober-netties? What the hell is coober-netties?”) — and there’s no way they’d have believed the other half. C# running on an iPhone? HA! ASP.NET on Linux? HAAA! And when you sat down an hour later and worked out how install a quantum computing simulator on macOS and then hack together a quantum entanglement demo using a free open-source code editor with support for an experiment language for writing quantum algorithms, they would not have believed for a second that Microsoft and .NET was the driving force behind all those amazing innovations.


It was one of the most positive and inspiring events I’ve been to in a while — and when I think of how many excellent conferences and meetups I’ve been to in the last few months, that’s really saying something. Huge thanks to Ian Russell and Dave Evans for putting it together, to ImpactHub for the venue, but most of all to everyone who came along. An event like this is only as good as the people who show up, and today was absolutely awesome. See you all at the next one.

Oh, and the session on ‘are microservices dead?’ Yeah. Nobody turned up. I guess that makes it official. RIP microservices. It’s been… emotional :)


LINKS

Various things that were shown, shared, discussed or mentioned:

No comments: