Improvements

[lbradstreet]

Add after-retry and after-complete to lifecycles example.
Have an example where they add submit-job (and maybe await-job-completion).
Add an example where we can manipulate catalogs/workflows/etc to show the power of data and manipulating data

[lbradstreet]

• Could've done with a high level overview about why you might want tasks to direct to multiple tasks (e.g. send out to a database as well as put something on queue).
• test 1-3 should have input-output descriptions.
• people get stuck on how params work. Might also be worth having a description about why (can modify the task data at run time, thus modify the params at run time).
• 4-1 should explain more about what lifecycles are. Also don't use identity for the function where you want to read the batch because then you could either use onyx.core/results or onyx.core/batch.
• Reduce reliance on capturing standard out - sucks because it's hard to add printlns
• Move flow conditions to before lifecycles, but add a basic lifecycle example early on as it's required to use any plugin.

[lbradstreet]

Instead of relying on capturing std-out, force users to swap!/reset! an atom from their lifecycle functions.

[jackdempsey]

The lifecycle examples with the core/async stuff, i.e.:

;; We get a little creative to make the core.async plugin easier
;; to use. We add our own core.async/id parameter to the lifecycle
;; and reference this later to obtain a shared reference to a channel
;; via memoization.

is a little unclear. At first this seems special to this example, but then you realize it's done elsewhere. Once you realize that you look for where it's used and how and why this seemingly low level thing needs to be done at all (and isn't automatically handled).

Any clarification or work on this piece would probably be great. Happy to help if I can once I understand things better :-)