July 2018 - Nick's .NET Travels

Debugging and Contributing to MvvmCross

As one of the maintainers of MvvmCross I quite often help triage issues that are raised on GitHub. Issues are a mixed blessing – it’s really important to us that the community logs any issues that they come across so that we have a mechanism to track and resolve them with each release. However, the flipside is that issues become a dumping ground for developers who use MvvmCross, have come across an issue that they want, or need, fixed – they log an issue and assume that other contributors, or the maintainers, will resolve the issue in a timely manner. Before I get into the body of this post which will drill into how to debug MvvmCross, as well as how to proactively contribute to MvvmCross, I want to reiterate the point that open source software, which includes MvvmCross, ships as-is. This means whatever expectations you have of it working, or working a specific way, or any expectations you have on how issues will get resolved, or how quickly they’ll get resolved, are wrong. The only things you really know for certain is that there is source code that’s available for anyone to access and contribute to, and there are some versions of the library that are available via NuGet so that they can be easily consumed by developers. Of course, with this said, the MvvmCross team, like other OSS, do strive to address issues as quickly as they can but we all have day jobs and lives that mean our contribution to OSS is limited. In this post I will be covering what you can do when submitting an issue to make it easy for someone to resolve your issue as quickly as possible.

Debugging with MvvmCross using Source Linking

Note: Source Linking doesn’t work with all platform. If you building a Xamarin.Forms application, it’s worth creating a UWP head just in case you need to debug an MvvmCross issue.

We often get issues raised, either via GitHub, Slack or Stack overflow, that start along the lines of “my application doesn’t start (or crashes) when I do …..” Because the issue often resides within a commercial product that the developer is working on, we don’t typically ask developers to share their solution. Instead what we typically ask is that they use the Playground sample that’s included with the MvvmCross source code and see if they can reproduce the issue.

However, before getting to this point, there is a step that can often help identify the cause of any issue, which is to setup and use source linking to step through the MvvmCross source code. Source Linking is a technique that’s been around for quite some time and is something that MvvmCross has setup to allow developers to easily step from their code, directly into the source code of MvvmCross without any complex setup process.

Getting source linking to work isn’t always as easy as it should be. There are a few websites that cover how to enable source linking (such as this post by Cameron Taggart) and they all basically say the same thing, which is that to enable source linking you need to check the Options >> Debugging >> Enable Source Link support checkbox in Visual Studio.

Update: Whilst you’ve in Options, you can uncheck the “Enable Just My Code” as this setting will mask any internal exceptions that may be thrown during start-up. In the current version (6.1.2) there is an issue whereby exceptions raised during the start-up sequence in the Setup class are swallowed, preventing start-up completing. This will manifest differently on each platform but for a Xamarin.Forms application it may result in a white-screen being displayed on start-up and never proceeding to the first page of the application. Unchecking this option will mean that VS will break each time an exception is raised, allowing you to see the cause of the exception.

However, what’s not covered is what do you do if this doesn’t work. I don’t have a definitive answer either but I did find that after I toggled a few of the options (eg Enable source server support) I was able to get source linking to work.

Once you’ve got Source Linking enabled, how do you use it? There are a couple of options but they amount to the same thing – you need to trigger some debugging action that will allow you to step through the source code. The first option, and one that will work if something is going wrong within MvvmCross that may be triggering an exception that’s handled/swallowed. In this case you’ll need to change your Exception Settings so that all Common Language Runtime Exceptions cause Visual Studio to Break. After changing this setting, if an exception is raised within MvvmCross, you’ll be able to step through the code at the point the Exception is raised.

The next option is to put a breakpoint in a method that overrides a method defined within MvvmCross. For example in the ProxyMvxApplication class, which inherits form MvxApplication, we can override the ReigsterSetup method and place a breakpoint in the method.

When the breakpoint is hit, you can see in the call stack that there are methods in the MvvmCross library.

Double-clicking on one of these lines will trigger the source code to be downloaded so that you can then step through the code. However, before the source code is downloaded, you may see the following prompt:

At this point you can opt in or out of source linking for this debugging session. Once you agree to Download Source and Continue Debugging, the cursor will be taken to the appropriate line of code and you can continue your debugging session.

As you can imagine, being able to step through the MvvmCross code means that you can try to identify what might be causing the issue you’re seeing. It’ll also mean that should you need to log a GitHub issue, you can provide us with a much more detailed description of what’s going on.

Contributing to MvvmCross

Contributing to OSS can happen at various levels, mainly depending on the amount of time someone has available and their inclination to help the community. For most developers their first contribution to MvvmCross will be in the form of a new issue. When submitting an issue, it’s important that you select the most appropriate issue template as a starting point.

As a community we welcome all feedback and encourage developers to raise an appropriate issue. Whilst we will respond to “Questions and Help” issues, we’d prefer it if you reached out via Slack or StackOverflow and actually talk to us!

It’s important to remember that in order to follow up on an issue we need as much information as possible. For example, information on how to reproduce an issue is invaluable. Providing a series of steps on how to reproduce an issue is a good starting point, and will help reduce the amount of time it takes to try to work out what’s going on. Alternatively, instead of just listing steps, a better way to demonstrate an issue is to use the Playground sample that’s included within the MvvmCross source code and reproduce the issue there. Once you’re able to reproduce the issue, you can submit a PR for review which may form the basis for resolving the issue.

Quick Steps for Resolving Issues:

In your solution: enable all exceptions in the Exception Settings window of Visual Studio – this will point out if an internal exception is occurring
In your solution: enable source linking where possible – this will allow you to step through MvvmCross source to try to work out what’s going wrong
In the Playground sample app in MvvmCross: attempt to reproduce the issue
Ask on Slack (or Stackoverflow) for assistance from the MvvmCross community
Raise an issue on the MvvmCross GitHub
Submit a PR with the changes to the Playground sample that demonstrates the issue

Note: When submitting a PR, please mark the PR as work in progress by putting [WIP] at the beginning of the title. This will prevent the PR from accidentally being merged. Also, please link the PR to the issue so that it’s easy to understand the purpose of the PR.

You might be asking why I’m asking that you submit a PR. It’s for two reasons:

1) I’m lazy – when I read your issue, it requires quite a bit of effort for me to reproduce the issue. If you provide a PR, I can simply pull your branch, run up the Playground and see the issue for my self. Not only that, I can also commit changes directly on your branch which means I can fix the issue straight away.

2) I want you to join the MvvmCross community – you only really join the community when you start contributing code to the project. By getting you to submit a PR, you’re breaking the ice and getting involved. The PR is a WIP so there’s no judgment from the community; in fact they’ll be impressed that you’re willing to get involved. We’ll often reach out directly to you and discuss the issue.

The upshot is that by submitting a PR you have a much higher chance of getting your issue looked at #NoGuarantees

Good-bye HipChat, and don’t let the door hit you on the way out!

It was interesting to see this week that Atlassian doomed the future of HipChat and its successor, Stride, with an aggressive wrap up schedule, with the services set to be discontinued on February 15, 2019. At Built to Roam, as a consulting company, we use a number of messaging tools including Messenger, Skype, Skype for Business / Microsoft Teams, Slack, HipChat and a few others. The upshot is that none of these tools do a great job of even their primary function (i.e. chat conversation between two or more parties), as I’ve posted about previously.

As a couple of posts have indicated, the messaging market has become over populated – for a while it felt like I was installing a new messaging app every second day. When Teams first came to the market, there was a lot of criticism aimed at it because it was a primitive offering in comparison to both Slack and HipChat but it’s rapid growth has started to put pressure on other players in the market. I think a rationalisation of the market was due, and I’m not sorry to see the back of HipChat. As one of the older products in the market, it never quite understood the need for users to belong to multiple organisations and to be able to switch between them.

There are some posts that are talking up the closure of HipChat/Stride as an attempt by Slack and Atlassian to team up in the fight against Microsoft Teams. So the question is, will this make a difference? Will it slow the growth of Microsoft Teams? Will it help Slack win over the corporate space?

Recently, Microsoft Teams announced a free tier, which was one of the things that held a lot of smaller companies and teams from using Microsoft Teams. This move in itself has weakened Slack’s position in the market. However, the true hook for Microsoft Teams and in my opinion the sole reason for its wide adoption (because let’s be honest, it’s far from being a great product!), is that it allows users to sign in using their Office 365 / Microsoft 365 account. In other words, if your company has made, or is moving, to Office 365, you can use your existing credentials to sign into Microsoft Teams. And of course, once you do, you can see and communicate with all the other users in your organisation. Can do you do this with Slack? The short answer is no. The long answer is yes but you need to do a bunch of stuff, including pay a ton of money for stuff that should be out of the box (seriously like what the? https://get.slack.help/hc/en-us/categories/200122103-Workspace-Administration#configure-access-security).

The ridiculous thing is that integration into Azure Active Directory (i.e. use Office 365 and Microsoft 365 credentials) is pretty straight forward. Is there something that Slack can do to get the jump on Microsoft Teams? Yes, provide out of the box support to sign in using either G-Suite or Office 365 credentials. In the future there will be two types of organisation, those that use Office 365, and those that don’t. Most of those in the latter group will probably use some form of G-Suite, so providing out of the box support for G-Suite should be on the radar of any enterprise software.

I know this post has gone on a bit but my last point is that I wish services would stop charging a premium for improving the security of their service. Integration with Azure Active Directory and G-Suite should be include in the cheapest tier of any offering. Why would you compromise the security of your service and the data of your users by not providing this.

Who broke my messaging app?

Prior to the iPhone the majority of applications written were for desktop based devices, whether they be for a Mac or PC, or in some cases both. With the iPhone we saw an entire industry of mobile app development opening up from being a very niche, specialist field, into a field full of imposters who claimed they knew how to build a mobile application, just because they were now able to build and publish an iPhone app. The first generation of iPhone applications weren’t anything special and in fact if you compared the sophistication of most iPhone applications from that era to those being written for other mobile platforms, they were definitely a case of “beauty over brains” – You have to hand it to Apple, as they really did make it easy to build high quality apps that looked great, even if they suffered from a lack of substance.

So hang on, how does this relate to my messaging app? Well, whilst the quality and complexity of mobile applications has increased, one thing that hasn’t is that they are all single window. Due to the screen real estate, mobile application have a single window. With the introduction iPad and other tablets, we saw the use of split screen and master-detail style layouts but even these were all contained within a single window. In the context of a messaging application, whether it be Skype, Messenger, WhatsApp, Twitter etc, the layout is all basically the same – you have a list of contacts and/or groups, you tap into a contact or group and are taken to a chat interface where you can message. In the case of devices with more screen, often the contacts/groups will be listed in a pane on the left or right of the current conversation but in both cases, if you have multiple conversations going, you have to switch between active conversations.

I have nothing against the single window view when working on a mobile device, and I think I’m ok with it on a tablet device but when we graduate to a laptop/desktop we enter a world of inefficiencies and poor user experience. My mobile comes with me virtually everywhere I go but when I’m working, or at home surfing the internet, I’m typically on my laptop (as an aside, I’m running the Surface Book 2 i7 15” and it’s an amazing bit of kit – highly recommend it!). I’ll often have a combination of Messenger, Skype, Teams, LinkedIn and Twitter open and will switch between conversations within each as required. However, typically at any given point in time I’ll be involved in one or two conversations, and it’s really annoying to have to keep switching between these conversations and whatever else I might be doing. For example I might be having a couple of DM conversations on Twitter but at the same time catching up with what’s been happening on my main feed – this means I have to keep switching between conversations, rather than being able to keep them both open at once.

Now I think you can see where I’m going with this – I run an operating system that I think most of you will have heard of…. Windows…. not “Window” but “Windows” <<- plural!!

Why is it that very few modern applications make use of multiple-windows? Well, it’s because of the iPhone – for the last decade app developers (and don’t forget all those app designers who can’t comprehend how an app can use multiple windows) have been building apps that only use one window because building iPhone applications was where the money was, and still is. It doesn’t help that Android apps are also predominately single window.

Let’s go back a few years an look at some user interfaces that have been discarded in favour of the now pathetic single-window messaging app experience:

Do you remember mIRC?

Whilst it was technically one window, it did support having multiple conversations open at the same time using child windows.

Related image

Do you remember MSN/Live Messenger?

It had multiple windows that you could open an arrange how you like. In fact you could force a window to stay on top of everything so it didn’t get obscured by other windows you might be switching between.

Image result for live messenger multi window

So where to from here?

Well the good news is that Windows app developers already have great multi-windowing support out of the box – you just need to build your application to take advantage of it (see https://docs.microsoft.com/en-us/windows/uwp/design/layout/show-multiple-views).

The bad news is that I doubt any of the major messaging services will take advantage of this in the short term, so we’re going to be stuck with a sub-optimal experience. Neither Skype and Teams, both Microsoft products, have any form of multi window support. Messenger still looks like it has been thrown together by some intern at Facebook, so I doubt we’ll see any innovation there. Twitter is a PWA, so I’m not sure how easy it will be for it to take advantage of multiple windows – having said that, as a PWA it can access any of the Windows 10 APIs, so there should be no reason why it can’t spawn additional windows.

Microsoft Whiteboard

I noticed the other day that Microsoft have added another app to the suite of apps that are available to Office 365 subscribers, Microsoft Whiteboard (https://products.office.com/en-us/microsoft-whiteboard/digital-whiteboard-app). On downloading it from the Store I was immediately impressed with the overall look and feel of the app – very professional and clearly showcases what can be done with the Windows platform.

Having said this, here are my immediate frustrations:

Where are the apps for other platforms?

Namely iOS, Android, Web, MacOS – it’s very transparent that this is a push to promote how great the inking experience is on Windows but for this to be a viable solution for businesses it needs to be available anywhere

Why upload PNGs?

Being a developer I immediately ran the app through Fiddler and what shocked me was that PNGs are being uploaded. I haven’t delved into how the synchronisation process works when multiple people are collaborating but I can’t imagine any scenario where uploading PNG is efficient. If you look at what other shared drawing experiences do (eg http://cosketch.com) there is no sharing of image, rather the line segments are sent back and forth.

Why isn’t this a Control

Again with my developer hat on, this shared whiteboard needs to be made available as an Office 365 control that developers can simply drop into their application in order to integrate a shared whiteboard experience. As more applications are built that tap into the Microsoft Graph and leverage the fact that users are connected with either an MSA or an Office 365 account, having rich component such as this would significantly cut development time and make it easier to build amazing applications.

Overall I’m impressed with Microsoft Whiteboard and hope that this is a sign of some of the great innovation that the Microsoft 365 platform will bring with it.