v3 roadmap plan

[bensheldon]

I wanted to write this out, mainly for my own thinking. And would love any feedback input on it.

The general plan that I have imagined in v3 is:

1. Extracting and centralizing the Scheduler/Notifier/Poller/CronManager instances inside of a global (e.g. GoodJob.manager). Currently they either are loose inside of the CLI, or within individual adapters. Queue and thread configuration is global, so there should only be one set of these in the process.
2. Extracting the majority of the logic from Adapter#enqueue, and moving that up to the global GoodJob.enqueue level (or maybe GoodJob::Job.enqueue or GoodJob::Execution.enqueue?). That reduces the size of the adapter, and unlocks...
3. Creating something like GoodJob.bulk_enqueue for enqueuing multiple jobs at once. This isn't possible while all of the enqueue logic lives within the Adapter (because an application/jobs can have multiple GoodJob adapters). That unlocks...
4. The cleaner interface for Batches with block enqueue (e.g. batch.jobs { SomeJob.perform_later; OtherJob.perform_later }

[bensheldon]

I think all of these have been accomplished:

• Extract GoodJob::Capsule #861
• Add bulk enqueue functionality #790

I didn't extract a central #enqueue method, but was able to accomplish the other issues without needing it.