Rockford Lhotka

 Thursday, March 22, 2018

I am pleased to announce the release of CSLA .NET version 4.7.100 with full support for netstandard 2.0 and .NET Core 2.

The packages are now in NuGet. Once some final updates to the samples are complete and merged into master I'll create a formal release tag/page on GitHub.

This release also includes some other very exciting capabilities, including:

  • #760 Add support for ASP.NET Core 2.0
  • #759 Add support for EF Core (EntityFrameworkCore)
  • #813 Major performance improvement when data binding to large object graphs thanks to keithdv
  • #795 Add Transactional attribute back into netstandard 2.0 code
  • #800 Changes to configuration so it is possible to configure CSLA without any web.config or app.config files (such as in .NET Core, etc.)
  • #496 Support ClaimsPrincipal via new CslaClaimsPrincipal type
  • #729 ApplicationContext now defaults to using AsyncLocal to maintain values on the current thread/context with help from j055 BREAKING CHANGE
  • #712 Support in-place deserialization of an object graph
  • #748 Major improvements to serialization via MobileFormatter thanks to jasonbock
  • #763 Update to samples thanks to tfreitasleal
  • #688 Get ApplicationContext.User authentication working with ASP.NET Core thanks to dazinator
  • #766 Update to use latest UWP libraries for Windows 10 Fall Creators Update BREAKING CHANGE
  • #790 BUG: Fix AmbiguousMatchException in data portal thanks to iherwald
  • #710 BUG: Fix ambiguous Save method issue thanks to rabidkitten

There are a couple known issues with this release:

  • #818 Xamarin projects using System.Data.SqlClient show a warning about this assembly's location
  • #794 UWP projects show warning PRI263: 0xdef01051 messages relative to CSLA resource strings
  • #822 There are two "sets" of CSLA packages/assemblies: one for full .NET 4.6.1+ and one for netstandard (including Xamarin, UWP, .NET Core, etc.) due to a type error between netstandard and full .NET
  • #703 Though netstandard supports BinaryFormatter, that is currently not an option from the CSLA configuration, and this needs to be addressed

Regarding the NuGet/assembly split noted in #822:

Right now there are two "families" of CSLA .NET packages in NuGet. One that supports full .NET and one that supports all other runtimes.

The full .NET family must be used for the following types of app:

  • Windows Forms
  • WPF
  • ASP.NET (other than Core)
  • Console apps (other than Core)
  • Azure Functions
  • Any other runtime hosted by full .NET 4.6.1+

The netstandard family must be used for the following types of app:

  • Xamrin
  • UWP
  • .NET Core
  • ASP.NET Core
  • Any other runtime not hosted by full .NET 4.6.1+

What this means for you is that if your n-tier app is 100% full .NET or full netstandard then you can live within one of those families. BUT if your server is full .NET and your client is Xamarin (for example) then your business library assemblies need to be compiled twice: once for full .NET and once for netstandard.

The Samples\ProjectTracker app shows how this is done by using a Shared Project to contain the code, and two concrete class library projects that compile the code for full .NET and netstandard respectively.

Thursday, March 22, 2018 1:38:26 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Thursday, March 15, 2018

There's some really wonderful work happening now that mono runs in webassembly. There's the Blazor project, and perhaps more exciting is an initiative that has Xamarin Forms running in webassembly.

Blazor sits on mono in webassembly and provides a Razor style UI framework. You write your .NET (C#) code to run behind an html/css UI created using Razor syntax.

Frank A. Krueger just published a different approach, where Xamarin Forms runs in the browser via webassembly, so your Xamarin Forms app (XAML + C#) runs on the client inside the browser. If you follow the above link, you'll find a nice walkthrough on how to use the dotnet command line to create a simple project and run it in a browser.

I started there, and thought it would be fun to try a more robust Xamarin Forms project using Visual Studio, XAML and code-behind. Pretty much my standard approach when building apps for iOS or Android.

Here are my steps (and the code is in GitHub as XFOouiDemo):

  1. Run through Frank's walkthrough - in my case on Ubuntu in Windows 10 WSL (Windows Subsystem for Linux), where I have dotnet installed, and of course Linux already has python - the result is that you should have XF running in your browser with a simple button that counts clicks
  2. Put the walkthrough directory on /mnt/c somewhere so the files are available to Windows as well as Linux - my path is /mnt/c/src, which is c:\src on the Windows side
  3. Open the csproj file in Visual Studio
  4. Add a NuGet reference to Xamarin.Forms (at this point you won't be able to do a dotnet build on Linux because XF apparently doesn't yet work correctly with .NET Core)
  5. Build the project - or maybe just restore NuGet packages
  6. Close and re-open VS (I found this stabilized things a lot)
  7. Add a new item to the project - specifically a Xamarin Forms ContentPage
  8. Edit the xaml to show your content, or not, as you choose
  9. Edit the Program.cs file as shown below
  10. Build the project in Visual Studio
  11. Switch back to your WSL window and run python to rehost your new code
  12. Again navigate to http://localhost:8000 in your browser to load your updated app, and you should see your new XAML-based content

The updated program.cs file is this:

using System;
using Ooui;
using Xamarin.Forms;

namespace MyFormsApp
  class Program
    static void Main(string[] args)
      // Initialize Xamarin.Forms

      // Create the UI
      var page = new Page1();

      // Publish a root element to be displayed
      UI.Publish("/", page.GetOouiElement());

Basically just replace all the manual creation of controls with an instance of your XAML page (in my case Page1.xaml).

What might bend your brain a little here, is that it might appear that I'm compiling the code on Windows and running it on Linux. But that is not what's happening. Remember that the code is running in your browser, not on Windows OR on Linux. So it doesn't matter whether you compile on Linux or Windows, the output of the compilation is wasm code that's downloaded to the browser where it is executed.

Thursday, March 15, 2018 6:58:04 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Monday, December 18, 2017

A couple years ago I left the Windows phone ecosystem. It was a hard thing to do, because Windows 10 on the phone is pretty wonderful, but the lack of apps made the platform undesirable.

In the end I switched to the iPhone 6s, after trying Android and being dissatisfied at the time.

About a week ago I switched to Android. The price tag on the iPhone X, coupled with ongoing frustrations with iOS and the lack of control made me give Android another try. And the fact that a lot of people told me that Android has improved a lot over the past couple years.

My new phone is a OnePlus 5T, and I'm on AT&T.

After about 10 days with the new phone I wanted to capture my initial thoughts, before they fade.


  • Cortana is my default assistant - oh the relief - Siri remains like Cortana's dumber older cousin, and it is really really nice to be back where I have a good assistant!
  • The Microsoft apps (Office, Outlook, OneDrive, launcher, lock screen, Cortana, etc.) all play together so well on Android because they aren't trying to work around Apple's silly rules - the experience is more pleasant and integrated.
  • All the apps I used on the iPhone are here on Android, so I lost no functionality by switching platforms.
  • I can use Waze as my default navigation app finally - this makes me so very happy!
  • The screen and performance and memory and battery life and charge speed of the OnePlus 5T are all excellent - a big step up from my iPhone.


  • Bluetooth doesn't work "right" between Waze (or Cortana) and my Ford Sync system (specifically my Sync system has to be listening to the BT stereo device to get any audio from Waze or Cortana, so I can't use Waze and listen to Sirius Octane - arg!!) This was not a problem on Windows or iOS.
  • Activity badges on app icons are far from reliable Not a problem with Windows live tiles or iOS.
  • Some apps on Android are still not as good as their iPhone equivalents; iPhone is clearly the highest priority, followed by Android and so Android users are sometimes second class citizens.
  • Contact management sucks on Android, just like on the iPhone. Nobody has figured out how to make contacts work seamlessly from Office 365, Google, and (well, except for Windows Phone 7, which nailed it - but then even Microsoft messed it up with Windows 10).

For me the pros have already outweighed the cons. The Bluetooth behavior is a serious pain though, almost enough to offset the pros - how do Android people live with this crappy BT experience????

But as I say, the pros - Cortana as default assistant and the smooth integration of all the Microsoft launcher, lock screen, and apps make Android the superior experience.

Cortana is a big thing for me. I limped by with Cortana on the iPhone, but the experience is much superior on Android. And I use Cortana on my Surface and desktop. And I have a HK Invoke, so I have Cortana in my home as well (so she can do all my home automation and play music mostly).

I've had people ask if it was hard to leave the Apple ecosystem. But I wasn't in the Apple ecosystem, I consciously chose to stay in the Microsoft ecosystem while using my iPhone, because I didn't want the vendor lock-in from Apple.

You might laugh at that, because I'm arguably locked in to Microsoft. But Microsoft has no "horse in the race" with mobile, so their ecosystem is the one place you can go to remain neutral between Apple and Google, and that's a powerful benefit.

My OnePlus 5T is a flagship level phone for half the price of Samsung's flagship or an iPhone X.

Avoiding vendor lock-in isn't a goal in and of itself (though people often think it is). The goal is to be able to switch to the best value hardware or OS or platform with minimal penalty because you aren't locked into some vendor's world.

Having switched from one major mobile platform to the other, with virtually no pain, I now feel very comfortable saying that the way to avoid lock-in from Apple or Google is to live in the Microsoft ecosystem 😃

Monday, December 18, 2017 11:38:41 AM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Wednesday, November 29, 2017

A while ago I got an Amazon Echo - the little hockey puck one. But I haven't been real pleased, for three primary reasons.

  1. It usually requires me to ask it the same thing twice
  2. It frequently gets disconnected from my network
  3. I'm a heavy Cortana user, and very few of the third party services supported by Cortana are also supported by Echo (no OpenTable, Wunderlist, cross-platform reminders, etc.)

So I took advantage of the recent holiday sales and ordered a Harmon Kardon Invoke, which is a Cortana based device.

As with most consumer devices post-Kindle, the Invoke comes in nice, friendly packaging.

(the pictures are rotated by MarkdownMonster - I don't know how or why, or how to fix it - sorry...)

The device itself feels solid, has some weight, and is attractive.

The swirling light on the top is pretty cool, and the physical ring at the top of the device turns to act as volume. The top of the device itself is touch sensitive, so you can tap it or tap-and-hold to interact. But really, most interaction is by voice.

Setting up the device was easy. When I plugged it in the blue light started swirling, and I used the Cortana app on my iPhone (which I already use all the time anyway) to initialize the Invoke. The whole process was pretty much automated, including connecting to my wifi network. The Invoke updated itself from the Internet once connected

Using Cortana with the Invoke is the same as with my Xbox or Surface. "Hey Cortana" and away you go! All the stuff I can do with Cortana on my iPhone or Surface seems to work the same on the Invoke.

What the Invoke does that seems special (as compared to Cortana via iPhone, Surface, or Xbox) include:

  1. Playing music via Spotify (one step during configuration is to identify your streaming music provider)
  2. Making voice calls to cell phones and land lines (the Invoke includes a year of free calling via Skype)
  3. Interacting with SmartThings based devices (and others too, but I have a ST hub, so that's what I'm using)

Unlike the Echo, Cortana has yet to need me to repeat anything, so that's a bonus right there!

Of course the Invoke sounds way better than the Echo Dot, but that's not a fair comparison. A friend of mine who has a full-sized Echo says that the Invoke sounds better than the Echo. That makes sense, given HK's reputation for audio.

The most important thing for me is that the Invoke, using Cortana, was instantly aware of, and integrated with, my existing reminders, to-do lists, calendars, package tracking, flight tracking, and everything else for which I already use Cortana.

Where really embracing the Amazon Echo would require me to use a whole other ecosystem, the Invoke just gave me access to the ecosystem I'm already using.

There's no question that the Echo has access to a lot more apps than Cortana. But it doesn't give me access to the apps I'm already using, and I guess that's the really thing for me - I don't want to migrate to a whole new to-do list app, stop using Foursquare and Yelp and OpenTable. I like those apps and services, and I'm used to them, and with Cortana I get to keep using them.

Wednesday, November 29, 2017 3:17:52 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Tuesday, November 7, 2017

When I worked in bio-medical manufacturing batch processing was a core element of our overall IT strategy. Not on the mainframe, but on the VAX.

The OpenVMS operating system, combined with VAX Clusters, had an amazingly powerful and flexible batch processing model, on top of which I wrote a relatively sophisticated scheduling engine to manage our nightly, month-end, and year-end batch processing.

We'd run dozens of batch processes each night, and many more for month/year-end. There were dependencies, and some jobs could or couldn't run in parallel with other jobs. My scheduler managed all that automatically, allowing users to add/remove requests for processing jobs and ensuing they fit into the schedule appropriately.

That basic robust batch processing capability is something I've missed ever since coming to Windows (so for 20+ years now), and I'm really thrilled to see the feature available as a first-class offering in Azure.

The fact that Azure also provides low-priority/lower-cost batch hosts is a nice bonus.

Now people are going to have to dig back in their memories, or start from scratch, and learn how to fully leverage batch processing once again!

Tuesday, November 7, 2017 2:55:47 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Tuesday, October 10, 2017

The CSLA framework is now officially 20 years old. Seriously, some version of CSLA has been out in the world for people to freely use since 1997 when my Professional VB5 Business Objects book came out with the code on a disc attached inside the back cover.

Although I didn't technically call it "open source" back then, because the more common terms were "freeware" and "shareware", and I considered CSLA to be "freeware". Those terms eventually became subsumed by what we now know as Open Source Software (OSS), and of course that's the term I use today.

I'm proud of the fact that I've been contributing to the OSS world for 20 years. Not too many open source projects last this long.

And more importantly, I'm humbled and amazed by the emails and feedback I've received over the years - both in terms of people using CSLA, and also people who've benefited from reading my books. And of course by the amazing contributions by colleagues and community members, several of whom I'll call out later in this article.

The following is a pretty long post, but I thought that it would be good to record the highlights of the past 20 years.

Why Open Source?

Lots of people have opinions and views and motivations about why they use and/or contribute to open source.

Personally, my motivation to contribute flows from being an early user of the concept. I very much believe that my career wouldn't have gone where it did, except for me having access to freeware, shareware, and open source software.

Back in 1989 (give or take) I was making my living being a DEC VAX programmer. And 1-2 times each year we'd get a couple big magnetic tapes (like in those very old movies) that contained tons and tons of freeware, shareware, and early open source (GNU and copyleft) stuff. I'd typically come in on a Saturday, download the index from the tapes and then selectively download and play with various tools.

Things like awk (really gawk) that I used to transform how our company processed reporting data. Or like uupc, which was a UseNet protocol implementation for the VAX that allowed us to get store-and-forward email service to the world over our 1200 baud modems.

I looked like a frickin' rockstar thanks to those tools!

Similarly, at the time, I was an Amiga fanatic. And the Fred Fish collection of software was like a lifeline to all that was cool and fun in the world of ... everything!

Yes, like everyone I've needed to make a living, support my family, save for the kids college and my retirement. But, and I know this sounds sappy, having built my career on the shoulders of some impressive people's contributions, I was on the lookout for a way to try and do the same for others.

CSLA COM or "Classic"

CSLA started out as the "Component-based, Scalable, Logical Architecture" in the world of COM. And I didn't originally intend for it to be a standalone framework. My original thought was that this code was an example of what people could do, and it fit into the bigger narrative of my book.

By the time I wrote the Professional VB6 Business Objects and Professional VB6 Distributed Objects books though, it was already becoming clear that I was mistaken. I phrase it that way on purpose, because by that time I was receiving emails with bug reports, feature suggestions, and other comments - where people were clearly taking the code and using it as a framework, not as an example.

Want to see that old code (and a picture of me from 20 years ago)? I put the old VB6 CSLA code in a GitHub repo for safe keeping.

Kevin Ford and I worked together on a project in 1996-97, and the two of us argued extensively about how things should work. Without his input, CSLA would not have been what it was, and Kevin has been providing input and contributions from then through today.


Toward the end of the 1990's I was seriously considering switching to the shiny new Java world. Primarily because I'd become increasingly frustrated with COM and DCOM, and yearned for the more object-oriented aspects of Java. Fortunately for me, I became aware of things like a new "COOL" language, and "COM+3". Technologies that ultimately became what we know as the .NET framework, VB.NET, and C#. So I stuck with the Microsoft world through that time.

Working with the earliest pre-releases of .NET I tried directly porting the COM code to .NET. That sort of worked, but clearly wasn't what I wanted, because it didn't leverage the new OO features available to me. I believe I wrote and rewrote the core of CSLA three different times into .NET, each time learning more and more about how .NET worked and what I could do with it.

The end result was CSLA .NET 1.0. Still very much a bare bones "framework" that I still personally considered as much an example of what to do as a framework someone could use out of the box. But by that point I had reconciled myself to the reality that people would be using it as-is, so I use XML comments and a short-lived, but very good, OSS tool called ndoc to create a help file for the framework.

And of course CSLA .NET came out concurrent with my Professional VB.NET Business Objects book. That was republished not long after under the title Expert One-On-One Visual Basic .NET Business Objects because my books were purchased from Wrox Press by APress. Not that this affected CSLA .NET, but it obviously affected me and my books - in a positive way, as APress was great to work with!

CSLA .NET 1.1 and C#

In fact, I wrote the Expert C# Business Objects book with APress. The port of CSLA .NET from VB.NET to C# was almost entirely done by a colleague of mine: Brant Estes. And I think he did it in just a few days - amazing work by an amazing developer!

Thus started the era of maintaining the "same codebase" twice in two different languages. You can see it in the v1-5-x branch - two versions of everything.

This is also about the time that the term "CSLA" stopped being an acronym and became just an unpronounceable word. The acronym no longer applied really, because .NET is object-oriented, not component-based. What I should have done is given it a cool name instead of some letters. Saying "see es el ay" is tedious. People were saying "the Lhotka framework", or "the coleslaw framework", or the "cuz law framework". But I didn't have that foresight, and after 20 years I doubt I'll change the branding now.

So now we're in 2002-2003, give or take, and the Internet as we know it today was really coming together, including the idea that OSS projects should have public source repos. So I set up a Subversion (svn) server and a web site with downloadable zip archives containing the code and samples. I still figured people should leverage the code, not the framework as a "product".


That changed in 2005 with CSLA .NET 2.0, which primarily supported .NET 2.0 and generics. And of course the new Expert C# 2005 Business Objects and Expert VB 2005 Business Objects books. But somewhere around this time is when I started releasing not just the code, but also pre-compiled binaries for people to leverage.


In 2007 Microsoft released .NET 3.0, including WPF, WCF, and WF. Supporting WCF meant decoupling the data portal from .NET Remoting, so people could use Remoting, asmx, or WCF as a transport protocol. Supporting WPF meant lots of alterations to data binding code, because WPF interacts with the data binding interfaces differently from Windows Forms. Very messy and hard to accommodate, so this was the point in the evolution of CSLA where I realized the primary benefit was how it mitigates all the differences in data binding, not the data portal.

Around this time was my first foray into self-publishing content via the web, with my CSLA .NET 2.1 Handbook and Using CSLA .NET 3.0 ebooks. Both published in VB and C# editions.

CSLA .NET 3.7 and Silverlight

2008 brought Silverlight 2, the first time .NET fragmented into two totally different platforms. Getting CSLA decoupled from full .NET, such that it could also support the Silverlight flavor of .NET, was a major undertaking. Thankfully CSLA had Magenic as a patron, and Magenic committed several full time developers for a period of months to work through all the issues necessary for CSLA to run on both full .NET and Silverlight.

2008 is also the year I published my last paper books: Expert C# 2008 Business Objects and Expert VB 2008 Business Objects. By this time I'd switched entirely to C# myself, and a colleague named Joe Fallon was responsible for getting the VB code and book out there.

Subsequent to 2008 the VB source code for the framework itself was dropped, and only the C# framework code continued forward. Obviously the framework supports all .NET languages, so it can be used by VB or C#, but the framework code itself is C# only from 2008 forward.

Justin Chase was instrumental in the port to Silverlight. In fact, there were no good ways to run unit tests in Silverlight, especially if you wanted the same tests to run in .NET. So Justin created UnitDriven to solve that problem.

Marimer LLC

In 2009 I created Marimer LLC as an official company to own CSLA .NET and all related code, content, and business operations. Prior to that point I'd been operating as what's called a Sole Proprietor, and having an LLC provides a level of legal protection. Not that I've ever needed that, thankfully.

The trigger for this was a very large company who wanted to use CSLA .NET, but even though it was open source they didn't want to use a product that didn't have a company behind it. Other people were covering the legal costs to create the LLC and ultimately it was to my benefit, so I saw no downside.


A couple years later, in 2010, came some of the biggest changes to CSLA .NET since the addition of generics.

This is also the point where I switched completely to self-published books and videos via the web with the Using CSLA 4 ebook.

The has been online with that content for seven years now. Wow!

The massive work to decouple from .NET and support Silverlight was far more work than what we did in .NET 4, but it was largely transparent to users of CSLA .NET (which is kind of the point after all). But the changes we made to CSLA .NET 4 were breaking changes for users of the framework.

I say we, because by this point in time CSLA had accumulated quite a number of contributors. A list of contributors since we switched to GitHub is on the CSLA .NET contributors page. I'm truly sorry I have no way to list all the people who contributed to the svn codebase, though the CSLA .NET Open Hub page has captured quite a lot of historical data and says 61 people have contributed.

But back to CSLA .NET 4 and the big changes. What did did in this version was add the basis for a robust rules engine. This was based on lots of community feedback from our early attempts with a rules engine starting in version 3.8 and the influence of David West through his Object Thinking book.

Building on this rules engine, Jonny Bekkum has evolved the engine substantially over the past several years, and today it is able to handle extremely complex scenarios. Best of all, the engine retains its relative simplicity for less complex scenarios and rules.

Beyond that, Jonny has been a major contributor to other parts of the framework, one of my most trusted advisors, and a stalwart member of the broader community over many, many years.

I also want to call out Tiago Freitas Leal, who's run a parallel OSS project called CslaCodeGenFork, has worked closely with Jonny to manage the community-based CslaContrib project, and has been an active contributor to the CSLA framework and the community. Tiago has been involved for almost as many years as Jonny.

This is also about the time that Jaans helped create the initial NuGet definitions necessary to get CSLA into NuGet. Most of his nuspec files remain intact, though perhaps evolved slightly over time.

Enter mono, mono for Android, and Windows Phone

In 2011 the open source mono port of .NET became much more relevant to me than it had been prior to that point. Primarily because the mono project embraced Android and iOS as well as Linux.

Jonny Bekkum leveraged the decoupling of CSLA .NET from full .NET to get a version running on mono, and based on that I tried to get it working on Android and iOS. Android was successful, but technical limitations of mono on iOS blocked our progress at that time.

We also leveraged the decoupling from full .NET in order to get a version of CSLA running on the nascent Windows Phone platform.

Turns out that all that work back in 2007 to support Silverlight was incredibly valuable going forward, even though Silverlight itself ultimately had no future. Because CSLA .NET had been largely decoupled from full .NET though, we've been able to get CSLA .NET running on over 11 different flavors/versions of .NET simultaneously - all from a single codebase!


In 2012 Micrsoft came out with the ill-fated Windows 8, and what I personally describe as the evolution of Silverlight: WinRT. Porting CSLA .NET to WinRT took some effort, mostly because Microsoft messed around with the way the .NET reflection API worked. Why they changed one of the fundamental building blocks of .NET itself is beyond me, and it caused a great deal of pain. But we worked through it.

The other big impact was that the XAML supported by WinRT was comparable to that from Silverlight 2. Which is to say it was pretty limited compared to WPF or Silverlight 5. Fortunately I was able to dust off a lot of old Silverlight 2 XAML code and use it to support WinRT. This goes to show the value of having good source control history over a period of years!

I think this is also about the time that we switched the source code repository from svn to git, and the server from my personal server to GitHub. Bringing all the svn history forward was not trivial. At all. It was my colleague Nermin Dibek who convinced me to switch to git, and Nick Schonning suffered through the pains necessary to convert the svn history into git so it was (mostly) preserved.

Enter Xamarin

By 2013 that mono for Android and iOS initiative had been canceled by its patron/owner Novell. Fortunately the project remained alive via a new company named Xamarin. Better still, Xamarin overcame some of the earlier iOS limitations, and so CSLA .NET was able to provide full support for Android and iOS via Xamarin.

From a historical perspective, it is also worth noting that this is about the time that, thanks to the confusion around the iPad and Windows 8, enterprises generally decided to stop building smart client apps at all, favoring to focus on HTML5 web apps and the newly coined "single page app" concept.

So supporting Xamarin was important from my viewpoint, because it felt like people might never again write a .NET client-side app otherwise. Today I'm not so pessimistic, primarily because today Xamarin is owned by Microsoft and supports cross-platform development across Windows 7, 8, 10, iOS, Android, Linux, MacOS, and more. Damned impressive!

Windows 10

In 2015 Microsoft released Windows 10 to replace Windows 8. Now Windows 8 was largely a commercial failure, but I believe it did one very important thing: it broke everyone's conception about what Windows looked like and how it worked.

Windows 10 brought back many of the best-loved features of Windows 7, while keeping and refining the bits of Windows 8 that were actually good. Personally I really enjoy Windows 10, more than I did Windows 7 (which I thought at the time was the best Windows ever).

Windows 10 brought with it an evolution of WinRT called UWP, or the Universal Windows Platform. UWP had an interesting quirk, in that it ran on a super-early implementation of yet another flavor of .NET: .NET Core. Getting CSLA .NET to run on this new UWP flavor of .NET wasn't too hard, primarily because it really is just an evolution of WinRT, which we already supported.

Support for UWP means that CSLA .NET supports native Windows 10 development, and also works for Xbox and Hololens development. I think that's pretty cool!

.NET Analyzers

Jason Bock has been a colleague and friend for many years, and his passion for compilers and low-level implementation details around .NET is hard to match. In 2015 Microsoft introduced this idea of .NET analyzers, which allow us to write components that runs within the .NET compilers and analyzes our code as we write and compile it.

Jason wrote a bunch of analyzers for CSLA .NET that help enforce the guidelines people should follow when using the framework. Prior to this point those guidelines were documented in my books and code samples, but there was nothing to help remind people what to do as they were coding. Jason's analyzers changed that entirely, providing real-time feedback and code refactorings directly within Visual Studio. Awesome!

MIT License

From the inception of CSLA until 2016 I'd been using a custom license based largely off the Apache license. And that never really caused any issue, but it did occasionally result in me having an email exchange with an organization's lawyer as they approved the license.

Switching to the MIT license is something, in retrospect, that I should have done much earlier. Now I don't get any such email exchanges, because everyone knows and understands the MIT license.

That's perhaps one of the lessons to learn here. I started doing all this back before "open source" was as formal or defined as it is today. And "little" things like licensing rarely get updated over time, because the focus is always on the code and other fun stuff. It is probably a good idea to review things like licensing more frequently over the lifetime of a project, just to make sure those elements remain up to date and relevant as times change.

.NET Core and NetStandard

The most recent changes to the .NET world have largely centered around the new .NET Core implementation of .NET, a version of .NET than is truly cross-platform. And something called netstandard, which is an API definition against which we can write code, and have that same compiled code work on multiple flavors of .NET.

CSLA .NET 4.7 supports netstandard 2.0, and that means a single Csla.dll works on all the platforms that conform to netstandard 2.0, including:

  • Full .NET 4.6.1+
  • Xamarin Android, iOS, and more
  • UWP starting in October 2017
  • .NET Core 2.0

Basically all the modern versions/flavors of .NET now support netstandard 2.0. And so does CSLA .NET 4.7.

So where we've been compiling the CSLA source code 11+ times - once for each flavor of .NET that's running out there, the future looks bright, in that hopefully, eventually, we'll be able to build it just one time.

Of course it'll be a while before that happens. Today CSLA still supports people as far back as .NET 4.0. But the future is bright, and today's messy reality of various versions/flavors of .NET and PCLs and everything else is clearly going to be replaced by netstandard.

Thank you

It has been a pretty incredible 20 years. I've met people (virtually and IRL) from all corners of the globe, and I've become good friends with many folks.

I've been able to continually improve and maintain the CSLA framework through the revenue generated by my books, training videos, and by having Magenic as a patron. Even with all that though, none of this would have been possible without the fantastic support from many thousands of people who've read my books and used CSLA, and the hundreds of people who've participated in the CSLA forums over the years, and the scores of people who've directly contributed to the CSLA framework and related projects.

As I wrote this article, I called out the names of a few people who've helped shape and create CSLA and the community. But there are other people in our small community that were and are also part of the narrative. Please know that I value each and every one of you as a colleague and friend.

Closer to home, the steadfast support from my wife and kids and close friends - really I don't have the words to express my gratitude and love for all of you. Much of my early work on CSLA and the books was done with my sons bouncing on my knees, sleeping in my lap as I typed, or making sure I got outside occasionally for some sunshine. And believe it or not, my wife and I are almost going to celebrate our 30th anniversary. So even with me disappearing to write for extended periods of time, we're still together and enjoying life.

CSLA .NET Futures

Where to from here?

I'm more optimistic about the future of .NET than I have been for many years.

.NET on the server is gaining new life and enthusiasm thanks to .NET Core and the ability to run your code on Windows, Linux, MacOS, in Docker containers, and I'm sure more to come.

.NET on the client might be on the rebound. I think as more people discover just how horrific Electron is as a development platform for desktops, and they are already backing off from Cordova on mobile, this will open the door for Xamarin and .NET to be viewed as a desirable alternative. You know, a technology that doesn't suck down memory and CPU for no good reason, and a technology that truly allows developers to leverage each device and platform when needed.

And then there's the great hope, which is web assembly (wasm). If this project continues to progress, and I think it will, browsers will no longer be a homogeneous runtime for JavaScript. They'll be much closer to a true operating system or runtime, supporting many languages, include C# and .NET. Just imagine going to a "web page" and having it run .NET code instead of JavaScript!

And where .NET goes, CSLA .NET goes too.

Tuesday, October 10, 2017 12:36:47 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Friday, September 8, 2017

ASP.NET Core only works with ClaimsPrincipal. Specifically, the http context from ASP.NET Core only accepts a ClaimsPrincipal instance, not the more general IPrincipal type.

Confusingly, the rest of the .NET world (full .NET, Xamarin, .NET Core, and netstandard2.0) still support IPrincipal. This makes ASP.NET Core an outlier, but an important one.

As a library author, I’m wondering if the consensus is that IPrincipal is dead, and that all principal objects should now subclass ClaimsPrincipal? Is this a new universal truth?

Specifically, should I run through all of CSLA .NET and in the netstandard2.0 version only support ClaimsPrincipal?

This would ultimately affect people building for Xamarin, full .NET, UWP, .NET Core, mono, as well as ASP.NET Core.

This would be a major breaking change for anyone trying to get existing .NET code (using CSLA) to run in any netstandard2.0 environment. The thing is, if you want to use ASP.NET Core you are kind of forced into that major breaking change anyway right?

My first reaction is NO – I shouldn’t make such a big change, because all the platforms not running in ASP.NET Core shouldn’t be forced to accept this burden just because ASP.NET decided to make a low-level breaking change by not supporting the IPrincipal type.

But I’m interested in hearing other people’s thoughts on this. What is the right answer?

Friday, September 8, 2017 3:16:28 PM (Central Standard Time, UTC-06:00)  #    Disclaimer

I was just reading this article about how to migrate from LastPass to 1Password.

I can't argue with what the author says in terms of LastPass having had some security issues. So I quickly checked to see if 1Password supported Windows 10.

It does not. No app in the store, no plug-in for the Edge browser.

Conversely, LastPass has an Edge browser plug-in and a (clunky-but-functional) app in the store.

Having a password vault and actually using a password vault aren't the same thing, and I'm pretty sure I wouldn't actually use one if it was a pain.

So at the moment LastPass wins, because they've put in the work to make it easy to use on my Win10 and iOS devices.

Their store app could be a lot better, but even a clunky app is infinitely better than no app at all.

Friday, September 8, 2017 2:55:01 PM (Central Standard Time, UTC-06:00)  #    Disclaimer
 Thursday, September 7, 2017

One of Magenic's largest business areas is QAT, and we have a serious focus on automated testing (including our open source MAQS testing framework).

The Software Test Professionals Conference (STPCon) is the leading event where test leadership, management and strategy converge.

I'm extremely pleased that this year's STPCon keynote speaker is Paul Grizzaffi) from Magenic.

Join us as Paul Grizzaffi explains responsible ways to approach automation, some of the knowledge we’ll need in order to be responsible, and shares insights about automation responsibility from his own career. Let’s allow history to remember our automation initiatives fondly instead of as Pyrrhic forays into irresponsibility.

Paul is also hosting a round table discussion on automation challenges.

Please join us in a round table discussion of attendee-provided automation challenges where we can share our thoughts and potential solutions to these challenges.

Troy Walsh, Magenic's practice lead for QAT, is also presenting at the conference. He'll talk about WinAppDriver vs Winium.

In this session, we will go hands on with WinAppDriver and Winum. We will dig into code and see how each tool works. We will also compare and contrast the tool features, usages and shortcomings.

Finally, Paul and Troy will team up to provide a demo of the open source MAQS framework.

We will demonstrate the Magenic Automation Quick Start framework, and its integration with CI/CD/CT workflows. MAQS is an open source package designed so that you can be running automated tests in minutes.

We are proud to be involved in STPCon, and hope you'll join us at the event!

Thursday, September 7, 2017 8:14:22 AM (Central Standard Time, UTC-06:00)  #    Disclaimer