Consider automatic deref

[christianromney]

Sometimes a complex data structure has something deref-able inside of it. This data is opaque to datawalk. It would be useful to automatically deref any node for which the predicate (partial satisfies? IDeref) is true when navigating to it with (w).

Also happy to contribute a PR if you feel this is something you'd accept.

[eggsyntax]

Terrific idea. I'll give that one a bit of thought to see if I see downsides, but unless something comes up, totally happy to make this change. It'll maybe take some consideration on how this ought to interact with backward, forward, and up.

Thanks!

[eggsyntax]

A few notes on this -- I'm experimenting with these options to try to see what feels like the cleanest solution.

• Ideal solution: 1) Works correctly with forward & backward 2) Works properly with up. 3) Works correctly with path
  • What does that even mean? A deref step means get-in will break anyway. 4) Seems natural to users
• Options:
  • Deref is an ordinary drill step 1, !2, 4 - breaks 3 (by default)
  • Deref is a special command 1, 2?, 3?, 4?
  • Deref is implicit, you never see atom Not sure yet exactly what this or the next option would mean
  • Deref is a printing affordance only, can't drill
• note that some derefables will cause problems, since they don't currently contain vals.
  • But deref optionally takes a timeout & a timeout-val; that should solve.
• Togglable setting for deref behavior?

I may start with a minimal solution so that users have at least some way of handling derefables.

[eggsyntax]

It also occurs to me that drill may ultimately be better as a protocol so that users can define for any type what it means to drill into it.

[eggsyntax]

Fixed in 0.1.11, now on clojars. Deref is an ordinary drill step. LMK how that works out for you :)

[christianromney]

Awesome, can't wait to try it out. Thanks!

[eggsyntax]

Bah. Crashes on cljs because there's no timeout version of deref there. Fixing.

[eggsyntax]

Fixed in 0.1.12.