Sagas implementation

[thangchung]

Consider could we use the https://github.com/dotnet-state-machine/stateless to implement the state-machine for the admin approval the article submission from the editor. Maybe we need to choose the workflow lib to make it work

[thangchung]

Read this for more clear about state machine, workflow and sagas https://medium.com/@roman01la/confusion-about-saga-pattern-bbaac56e622#.7x70mb8q3
Another collection of articles about Sagas at https://github.com/thangchung/magazine-website-akka/wiki/Saga-research

[thangchung]

stateless + quartz + topshelf

We can use Quartz (https://github.com/quartznet/quartznet) to schedule the job (run forever) and inside that job, we can trigger the stateless's state machine to get it work. Not sure it works or not? Need to check
https://stackoverflow.com/questions/25299662/executing-multiple-quartz-net-jobs-with-topshelf
I think Topself will boot the Service to run the persistence actor, inside that actor will get events from event store to re-init the state of the Aggregate of entity, then we can instance the stateless's state machine for run in every 1 minute or more by using Quartz

akka FSM + quartz + topshelf

The best way to do here is akka state machine with Quatz

• http://getakka.net/docs/FSM
• http://getakka.net/docs/persistence/persistent-fsm
• https://petabridge.com/blog/akka-actors-finite-state-machines-switchable-behavior/
• https://petabridge.com/blog/akkadotnet-async-actors-using-pipeto/
• http://doc.akka.io/docs/akka/snapshot/scala/fsm.html

[thangchung]

Really nice example using stateless at https://sachabarbs.wordpress.com/2013/05/16/simple-but-nice-state-machine/

[thangchung]

A lot of references in case we need to learn more
http://blog.jonathanoliver.com/cqrs-sagas-with-event-sourcing-part-i-of-ii/
https://abdullin.com/lokad-cqrs-retrospective/
https://msdn.microsoft.com/en-us/magazine/mt238399.aspx?f=255&MSPPError=-2147217396
https://www.slideshare.net/PieterKoornhof/event-sourcing-and-why-it-is-a-good-choice (https://github.com/SneakyPeet/EasyEventSourcing)

// in the FE
paldepind/functional-frontend-architecture#20 (comment)
https://blog.javascripting.com/2015/08/12/reduce-your-side-effects/

[thangchung]

https://github.com/vrcod/akka-sample-persistent-fsm/blob/master/src/main/scala/PersistentFSMExample.scala

[thangchung]

The next refactor should be as

1. Separate the process manager into another service, maybe ProcessManager service
2. Make current CategoryService builds the actor selection that points to the ProcessManager service
3. Then each request in CategoryService will trigger the command to the ProcessManager service
4. ProcessManager service will take care of inter-communication with side-effect services like email, pdf... services to finish its works and persistence its state into the storage via akka's PersistenceFSM

[thangchung]

Good for reference at http://vasters.com/archive/Sagas.html