reF#actoring: rewriting an actor in F#

The article explains why the team rewrote an Akka.Quartz.Actor-based scheduling actor in F#. Their goal was to keep the scheduling behavior they needed while removing client dependence on Quartz.NET details and on a direct Quartz.NET reference.

It compares the original C# project with the F# version and shows that discriminated unions, a functional actor loop, and a single-file layout made the code smaller and simpler. The article also notes that the F# version was tested with Akkling-style validation helpers and that the rewrite cut the code size by about 60%.

Reading notes

#

• The team used Akka.NET as the actor framework in a media distribution engine at NRK.
• They tried an open-source scheduling actor and then chose to port it to F# for their own needs.
• Akka.Quartz.Actor already existed, and it had a NuGet package by the time of writing.
• The main concern was that clients still had to reference Quartz.NET and its trigger types.
• The F# version introduces a JobSchedule discriminated union for the scheduling DSL.
• createTriggerBuilder adapts the union cases to Quartz.NET trigger builder calls.
• The original C# command and event definitions were split across many files and much longer.
• The F# command and event definitions were condensed into a single file.
• The actor was rewritten as an F# function with a recursive loop inside actor { ... }.
• The F# actor starts and shuts down the Quartz scheduler and handles CreateJob and RemoveJob messages.
• A private QuartzJob type passes the message and actor reference through Quartz JobDataMap.
• The resulting code lives in one file called Scheduler.fs and is 106 lines long.
• The article says the rewrite reduced code size by about 60%.
• For testing, the author used helpers from Akkling to write F#-style actor tests.
• The conclusion argues that F# improved compactness and readability for this actor-based code.