General errors #177
General errors #177
Conversation
CHG: Killing a Fiber during the WAITING state defers the kill action.
CHG: Applied Nate's code cleanup suggestions.
src/Effect/Aff.purs
Outdated
|
|
||
| -- | A monomorphic version of `try` that can map the error type. Catches thrown | ||
| -- | errors and lifts them into an `Either`. | ||
| attempt ∷ ∀ e1 e2 a. Aff e1 a → Aff e2 (Either e1 a) |
There was a problem hiding this comment.
There should be Void instead of e2 I suppose
There was a problem hiding this comment.
Aff e a does not mean an error of type e is definitely thrown; it means an error of type e may be thrown.
There was a problem hiding this comment.
Yes but if you do attempt on some Aff computation it must eliminate all possible errors, which means resulting Aff must throw nothing. If you stick Void there it would mean Void may be throw but as there is no value of type Void it means nothing can be thrown.
There was a problem hiding this comment.
That would prevent you from attempting inside an exiting Aff e1, unless you lmap absurd (attempt ...) though, which would be a bit annoying.
There was a problem hiding this comment.
In general you only want to use Aff Void a when you want to demand no exceptions from the user (taking an Aff as an argument), and use Aff void a when you are just not using exceptions (returning an Aff), since it means you can unify with any Aff context.
backwards-compatibility API in its place.
ADD: #! operator as an alias for 'flip lmap' ADD: absurdL/absurdR for free error conversions from Void. CHG: A synchronous throw in makeAff now panics (used to throw) CHG: Made killFiber more general. CHG: Cancelers are no longer allowed to throw. CHG: Bracket finalisers are no longer allowed to throw. FIX: Handle parallel Fibers that resolve synchronously (fixes cannot call 'run' on undefined)
WIP CHG: AffResult replaces Either
…rescript-aff into general_errors
|
By the time I achieve what I want it is too far incompatible with the original Aff. I am instead working on a new library that builds from the lessons of Aff. I will retract both these PRs. |
Includes #169.
I expect to receive a lot of backlash over breaking backwards compatibility. What I hope is that this can start discussion about how to move forward.