New plans and stuff

So after finally getting MapGuide Open Source 4.0 out the door, I took a self-imposed hiatus from all things mapping/GIS related for several months, permanently moved from Windows to Linux as my daily driver OS just in time before the end of Windows 10 support, and also to mentally recharge and savor the relief of having this major burden (of releasing MGOS 4.0) being finally lifted off of my back.

I now return with a renewed vigor and some rough roadmaps for things going forward in MapGuide and my other various projects. Part of that renewed vigor is due to the advent of ...

GitHub Copilot

In the past few months in my day job, I have been exposed to GitHub Copilot and it has changed the way I build and ship software, some changes bad, some changes good. Say what you will about AI (or AI-generated code/content) in general, but GitHub Copilot (or any other AI coding assistant) has ultimately been a net positive for me.

What separates how I use AI coding assistants from most depictions of "vibe coding" is that I know the technical and architectural fundamentals of what I am actually after. So I know when GH Copilot is generating what I'm after and when it's generating garbage, and knowing the right prompts to guide it back on track if it starts going off the rails, or know when to cut my losses if the situation is un-salvageable.

So with several months of GH Copilot usage at work, I have been thoroughly convinced that I should get a GH Copilot pro subscription for my own personal use. So last month, I finally bought a GH Copilot pro subscription and ... proceeded to blow my monthly allowance of usage credits in 2 weeks! 😂

But in those 2 weeks, I was able to make some major progress in mapguide-react-layout, knocking off some long standing technical debt and feature requests, some of which you'll see in future dev diary updates on this blog. The productivity gains were massive and turnaround times were quick! A new month has rolled over and with that, a reset of my monthly allowance and I have since learned to judiciously use GH Copilot in a less wasteful manner.

So, now armed with GH Copilot, I have a rough roadmap of things I want to achieve in my various projects, which are all outlined below. So let's start with MapGuide.

MapGuide Open Source

The next release of MapGuide Open Source will be 4.0.1. It will be a bug fix release and include any updated web tier components and upstream FDO fixes since the 4.0 release.

Since permanently moving to Linux as my daily driver OS and understanding that I will still need to produce Windows builds of MapGuide and FDO, I've been re-establishing my windows dev environments for MapGuide/FDO inside a virtual machine and reinstalling all the necessary dev tools. 

One surprise that caught me off-guard was installing Visual Studio 2026 Community Edition. Even though MSVC 2019 is the standardized windows compiler for building MapGuide/FDO on Windows, we can install older compiler workloads on newer releases of Visual Studio. So as part of reinstalling all the required dev tools, I thought it would be a simple case of installing VS 2026 with the MSVC 2019 compiler workloads and everything should be all good, right?

Well, near the tail end of my first MapGuide windows build inside this new VM, I hit a snag on the Windows Installer portion. It turns out our WiX 3.x installer projects are no longer supported in Visual Studio 2026! I was not going waste time and disk space to install VS 2022, so I looked at what options we have for Visual Studio 2026. The answer was to migrate our installer projects to WiX 6.x.

Before GH Copilot, the migration process would've been tedious, having to read migration docs, assess the impact of any breaking changes, etc, etc. But with GH Copilot, I was able to migrate and iterate rapidly towards a WiX 6.x windows installer project that successfully builds and produces a MSI installer.

Now here comes the tedious part that GH Copilot cannot help me with. I still have to manually test this Windows installer, making sure all the feature toggles, custom actions, registry and start menu registrations still work as before and I expect this to take several weeks.

Once this updated Windows installer has been verified as working, there'll be a 1-2 week window of incorporating any bug fixes and updated web tier components before putting out the 4.0.1 release.

mapguide-rest

We will start wrapping up various loose ends on this project. The next release will be 1.0 RC7 and will:

• Drop support for versions of MapGuide Open Source older than 4.0.
• Represent mapguide-rest as being feature complete. RC7 to final will be bug fixes only.

I expect GH Copilot to pull heavy duty in helping me knock off all of these long standing items.

mapguide-react-layout

The next release will be 0.15. It will have some exciting new features that you'll see in future dev diary entries on this blog, but the main feature will be to fully decouple ourselves from the Blueprint UI toolkit. Blueprint gave us a nice cohesive set of UI building blocks, but our production bundle sizes have paid a price for this convenience, not to mention that integrating this viewer into other projects that use other UI libraries is problematic as Blueprint is always included, whether you like it or not.

As part of this dev cycle, I've been taking inventory of everything in Blueprint that we actually use and refactoring their usages through a layer of indirection of "abstract UI components" so that we can supply an alternative implementation that is more bare metal and whose appearance and styling can be customized through good ol' CSS.

0.15 will be on the horizon upon the completion of this Blueprint replacement. GH Copilot has paying massive dividends in terms of progress and momentum on this particular project, so I am hoping for a fast turnaround on these remaining items.

MapGuide Maestro

Finally we come to MapGuide Maestro. I will finally bite the bullet and put an end to this endless series of 6.0mSomeNumber releases by putting a final MapGuide Maestro 6.0 release. This 6.0 final release represents the end of MapGuide Maestro in its current form as a Windows Forms based MapGuide authoring and administration tool.

After the 6.0 final release is out. MapGuide Maestro will be rebuilt from the ground-up as a true cross platform MapGuide authoring and administration tool. It will be built on modern .net (and all of its patterns and practices) and our ability to have a true cross-platform UI will be achieved through the use of Avalonia as our UI toolkit. 

A little experiment some years ago plus experience from building/maintaining a separate personal project that uses Avalonia for the UI (not yet announced/revealed on this blog, maybe I will someday) and once again armed with GH Copilot, I have great confidence in pulling off a ground-up rewrite.

And that is a peek into what's in store in the near future. Exciting times are ahead!