My Spectre Console Foo needs a bit of work, but the test runner is actually testing now with live changes 
I'm finally close enough to Perla v1...
slow and steady
#dotnet #fsharp #webdev #testrunner #testing
#fsharp
2 posts1 participant0 posts today
There we go, it has been quite tense for me to get back and support somewhat of a test runner for JS files within the tool but looks like we're headed to a good place.
Last time the problem was the bundle size massively increased when I included playwright in the tool 
The perla docs website is now up!
The "Real World" Fable sample has been updated to the latest perla preview, links in the following tweets!
https://github.com/AngelMunoz/real-world-fable
You can check the commit history, upgrading this was a breeze just bumping package versions!
https://angelmunoz.github.io/Perla/
The docs website is still a work in progress, however feel free to ping me if something feels weird or if you get lost!
GitHubGitHub - AngelMunoz/real-world-fable: A Single Page Application written in F# using Fable and Elmish.A Single Page Application written in F# using Fable and Elmish. - AngelMunoz/real-world-fable
Perla v1.0.0-beta-028 is out
I keep fixing stuff from here and there so, if you're using it be sure to keep an eye to the releases tab often in case there's new goodies!
GitHubReleases · AngelMunoz/PerlaA cross-platform web standards-first SPA dev server - AngelMunoz/Perla
We just shipped Visual Studio adapter 3.1.3.
This is a bug fix release to address a failure case when running xUnit.net v2 tests (which was introduced in 3.1.0).
xunit.netVisual Studio adapter 3.1.3 [2025 July 16] | xUnit.net
Another Perla release! v1.0.0-beta-026 is out on NuGet and the releases page
This mini release focused on ensuring your package installs can be downloaded locally and built without issues
Check it with the new basic local template
perla new -t blocal my-project && cd my-project
Example with preact dependencies:
perla add preact htm/preact @ preact/signals
Add your code then run "perla serve" or "perla build --preview"
https://github.com/AngelMunoz/Perla/releases/tag/v1.0.0-beta-026
GitHubRelease v1.0.0-beta-026 · AngelMunoz/PerlaWhat's Changed
Fixes
PkgManager: Fixed directory creation before symlink operations, preventing runtime errors during package management.
ProviderOps: Corrected package name extraction for flat st...
Replied to Urs Enzler 
@UrsEnzler I’ve come to believe from managing a bunch of complex, internal biotech code bases, that the specification is half of the problem so nowadays, I would prefer people dwell on the ticket for longer before they even code. Data structures are the most important thing to get right because it’s like a lake you build a village around. It’s awkward to move it later. Finally, the AI towing almost effortlessly builds the UI prototypes. There is such a ton of boilerplate standard UI code out there that following what everyone does is really the right thing I’ve had less success using AI to refactor complex backend #Fsharp, but I think it’s the complexity of the code not the language. It’s pretty good at spitting out standard implementations of common data handling patterns. E.g. build me an interval tree.
Today, this thought crossed my mind:
Claude Code appears to be the current pinnacle of AI-supported software development. In the videos I've seen, considerable emphasis is placed on the importance of a detailed and precise specification. The better the spec, the better the outcome.
That's not surprising, but it makes me think about a choice here:
1) Iterate on a precise specification
2) Iterate on concise code directly
I know this space is already won by vite/esbuild however I think there's a space for web-first incremental approach to SPA development
This is the first pre-release in a couple of years however I think things will flow better from now on
https://github.com/AngelMunoz/Perla/releases/tag/v1.0.0-beta-025
GitHubRelease v1.0.0-beta-025 · AngelMunoz/PerlaWe're somewhat back
This is the first beta of perla v1 in years and I think it was for the best. I won't dive into details but we're closer than ever to release v1.
There's going to be some rough p...
Look ma' no esbuild, no cdn, and somewhat editor tooling support! (you wouldn't believe how much the tooling "just works" when your dependencies are stored in a "node_modules" directory)
We just shipped Core Framework v3 3.0.0, Analyzers 1.23.0, and Visual Studio adapter 3.1.2.
Check the release notes for breaking changes, new features, and bugs fixed.
https://xunit.net/releases/v3/3.0.0
https://xunit.net/releases/analyzers/1.23.0
https://xunit.net/releases/visualstudio/3.1.2
xunit.netCore Framework v3 3.0.0 [2025 July 13] | xUnit.net
I guess we're at that point where it is clear that LLM are useful enough for some tasks, if you're not convinced yet for some reason let me share you a conversation between me and GH copilot agent
I tasked it to reimplement the Perla's dev server in suaveio
You can see where it was choking and I came in to do the manual work, however the rest was made by it in a few minutes at most
https://gist.github.com/AngelMunoz/47065f6603871ee64b238f7a23674c67
GistA GH Copilot suave implementation of the existing aspnet server that backs the Perla Dev Server. The overall generation of this file was a few mintues with a couple of hours of mine refining some details around the observable utilities as well as some of the handlersA GH Copilot suave implementation of the existing aspnet server that backs the Perla Dev Server. The overall generation of this file was a few mintues with a couple of hours of mine refining some d...
A small release is out for JDeck.
This release adds functions to decode F# Maps and BCL Dictionaries which were missing from the last bit (link to release in the next tweet)
https://github.com/AngelMunoz/JDeck/releases/tag/v1.0.0-rc-002
GitHubRelease v1.0.0-rc-002 · AngelMunoz/JDeck[1.0.0-rc-002] - 2025-07-11
Added
New Decode.map and Decode.dict functions for decoding Maps and Dictionaries
Enhanced Optional module to handle decoding of Maps and Dictionaries with null checks
...
This was a hell of a week
There were a LOT of changes to Perla, a massive refactor due in order to keep maintaining the project in a much simpler way.
Tests themselves need updates, we might have regressed a little bit and the test command still needs the same refactoring
However, the refactoring that took place also gave up some cool benefits like download sources support (e.g. local dependencies directory), and adaptive "live" configuration among others
https://github.com/AngelMunoz/Perla/pull/132
GitHubRefactor package management and enhance project structure by AngelMunoz · Pull Request #132 · AngelMunoz/Perla
We just shipped a new prerelease build of the core framework (3.0.0-pre.40).
There are no new breaking changes, so this will not reset our release clock for 3.0.0 (which should be in about a week).
xUnit.netRelease Notes for v3 3.0.0-pre.40 [2025 July 6]Documentation site for the xUnit.net unit testing framework
Continued thread
Just a reminder that we only have roughly 10 more days of before we ship 3.0. If you've been putting off validating your tests and/or extensions with the latest prerelease, your time is running low...
The State of Rx.NET in 2025
https://www.youtube.com/watch?v=SZpmj0fSfFI
tl;dw:
- Still living in dependency hell. They won't release Rx7.0 before solving those issues first. Everything else has lower priority.
- AsyncRx.NET will be developed over time.
- Ix.NET will not be developed as a part of the Rx project anymore. All LINQ to IAsyncEnumerable stuff will be moved to the core dotnet libraries.
While it has its own issues, there are several good reasons why my favourite #ProgrammingLanguage in the real world is #C.¹
Why not #Go?
Because it's from #Google.
Why not #Csharp or #Fsharp?
#Microsoft.
Why not #Rust or #Zig?
#LLVM (aka #Apple & friends).
Ultimately, most of languages I avoid like the plague are controlled by #BigTech one way or another.
C is simple enough to get several alternative compilers based on useful standards.² ³
¹ In theory I still prefer #Oberon07, but when I want to code something useful I still use C instead to lower the entry barrier for other devs, because there are too many incompatible implementations of the compiler and "standard" library.
² Ok, #Python, #Scheme and #Lua have similar qualities, but for the tools I write I usually prefer binary executables with no runtime.
³ No, #C++ is not an option.
Why not #Go?
Because it's from #Google.
Why not #Csharp or #Fsharp?
#Microsoft.
Why not #Rust or #Zig?
#LLVM (aka #Apple & friends).
Ultimately, most of languages I avoid like the plague are controlled by #BigTech one way or another.
C is simple enough to get several alternative compilers based on useful standards.² ³
¹ In theory I still prefer #Oberon07, but when I want to code something useful I still use C instead to lower the entry barrier for other devs, because there are too many incompatible implementations of the compiler and "standard" library.
² Ok, #Python, #Scheme and #Lua have similar qualities, but for the tools I write I usually prefer binary executables with no runtime.
³ No, #C++ is not an option.
harmful.cat-v.orgBjarne Stroustrup: "I Did It For You All..."