Wolverine 3.0 is Live!

Just as the title says, Wolverine 3.0 is live and published to Nuget! I believe that this release addresses some of Wolverine’s prior weaknesses and adds some powerful new features requested by our users. The journey for Wolverine right now is to be the singular most effective set of tooling for building robust, maintainable, and testable server side code in the .NET ecosystem. If you’re wondering about the value proposition of Wolverine as any combination of mediator, in process message bus, asynchronous messaging framework, or alternative HTTP web service framework, it’s that Wolverine will help you be successful with substantially less code because Wolverine helps you much more to simplify the code inside of message handlers or HTTP endpoint methods than other comparable .NET tooling.

Enough of the salesmanship, before I go any farther, let me thank quite a few folks for their contributions to Wolverine:

Babu Annamalai
JT for all his work on Rabbit MQ for this release and a whole host of other contributions to the “Critter Stack” including leveling us up on Discord usage
Jesse for making quite a few suggestions that wound up being usability improvements
Haefele for his contributions
Erik Shafer for helping with project communications
JasperFx Software‘s clients across the globe for making it possible for me to work on the “Critter Stack” and push it forward (a lot of features and functionality in this release were built at the behest of JasperFx clients)
And finally, even though this doesn’t show up in GitHub contributor numbers sometimes, everyone who has taken the time to write up actionable bug reports or feature requests. That is an absolutely invaluable element of successful OSS community projects

Alright, more lists! Here’s some relevant links:

The major new features or changes in this release are:

Wolverine is no longer directly coupled to Lamar and can now used with at least ServiceProvider and theoretically any other IoC tool that conforms to the .NET DI standards — but I’d highly recommend that you stick to the well lit paths of ServiceProvider or Lamar. Not that many people cared, but the ones who did cared about this a lot
You can now bootstrap Wolverine with HostApplicationBuilder or any .NET bootstrapper that supports IServiceCollection some how, some way. Wolverine is no longer limited to only IHostBuilder
Wolverine’s leadership election and node assignment subsystem got a pretty substantial overhaul. The result is much simpler code and far, far better behavior and reliability. This was arguably the biggest weakness of Wolverine < 3.0
There’s a new transport option for Apache Pulsar (actually really old code, but released to Nuget now)
Batch message processing
“Sticky” message handling when you need to handle a single message type in multiple handlers with “sticky” assignments to particular queues or listeners.
An options for RavenDb persistence including the transactional inbox/outbox, scheduled messaging, and saga persistence
Additions to the Rabbit MQ support including the ability to use header exchanges
Lightweight saga storage for either PostgreSQL or SQL Server that works without either Marten or EF Core

And plenty of small “reduce paper cuts and repetitive code” changes here and there. The documentation website also got some review and refinement as well.

What’s next, because there’s always a next…

There will be bug reports, and we’ll try to deal with them as quickly. There’s a GCP PubSub transport option brewing in the community that may hit soon. It’s somewhat likely there will be a CosmosDb integration for Wolverine message storage, sagas, and scheduled messages this year. There were some last minute scope cuts for productivity that maybe gets addressed with follow up releases to Wolverine 3.0, but more likely in 4.0.

Mostly though, Wolverine 3.0 might be somewhat short lived as Wolverine 4.0 work (and Marten 8) will hopefully start as early as next week as the “Critter Stack” community and JasperFx Software tries to implement what I’ve been calling the “Critter Stack 2025” goals heading into 1st quarter 2025.

I’m logging off for the rest of the night (at least from work), and I know there’ll be a list of questions or problems in the morning (the joy of being 5-7 hours behind most of your users and clients), but for now: