`handleErrorWith` ?

# arrow

[pakoito]

handleErrorWith ?

[melatonina]

I'll check that out. Thanks

[pakoito]

your implementation can be done (less optimally) using swap().flatMap(block)

[melatonina]

Looks like handleErrorWith() preserves the same type of failure (the left type)

[pakoito]

ye, you're correct

[melatonina]

I should use swap().flatMap(block).swap(). But then I'd lose the inlining, and I need that because I'm running suspend functions inside that block.

[pakoito]

oooooh

[pakoito]

cheeky

[pakoito]

flatMapLeft is a good name for this. I'm trying to remember why we didn't allow to change the error type on flatMap

[pakoito]

I'm sure it's because the MonadError instance is fixed to one error type

[pakoito]

which..makes sense

[melatonina]

This is the place where I'm using this: https://gist.github.com/mel4tonin4/9c4a1fec2f3ddec401552823dece1a4c

[melatonina]

processFlow is a builder of ProcessFlow which is something like a kotlin Flow emitting progress events, followed by a terminal event containing a result. The ``as`` infix operator embeds a ProcessFlow in the current ProcessFlow. ProcessFlow is actually a wrapper around a regular Flow, and that's where the need for using suspend functions comes from.

[melatonina]

Thanks, Paco!

[pakoito]

I would split the operation in two

[pakoito]

the fold inside flatMapLeft could happen outside

[pakoito]

where you have access to suspend

[pakoito]

doNotCompute? StoredEvaluationProblem.ComputationSkippedAsRequested.left() : computeFlow()

[pakoito]

hmmm

[pakoito]

meh I'm overthing it

[melatonina]

Using these 3 extensions methods https://gist.github.com/mel4tonin4/7472822197456889c2fd889b552431db I reduced the code to https://gist.github.com/mel4tonin4/10434d0d6f83773c57c9a374f558c83d

[melatonina]

I didn't need the `Option<T>.`if`` after all: https://gist.github.com/mel4tonin4/a4257b83a4f609474a78471b21639ad2.
This is the best I can do without using metaprogramming, I guess. If I keep using this ProcessFlow , I'll check arrow-meta and see if I can remove some boilerplate code (not visible here) and simplify the syntax even further.
Anyway, thanks again!