Stream: project-error-handling

Topic: Bidirectional control flow


view this post on Zulip Jubilee (Oct 16 2020 at 18:37):

I found a cool paper that seemed relevant for discussing error handling! https://www.cs.cornell.edu/andru/papers/ufo/

view this post on Zulip Mario Carneiro (Oct 16 2020 at 18:58):

Gosh, I feel like the algebraic effects literature is rediscovering call/cc all over again. At some point someone is going to point out that complex control flow is complex

view this post on Zulip Jubilee (Oct 16 2020 at 19:03):

In fairness, Rust is definitely an exercise in applying 40 year old technology, so while I agree, I also can't really hold that against them. :^)

view this post on Zulip oliver (Oct 16 2020 at 19:32):

Coq I think could certainly be described as 'comprehensive'

view this post on Zulip Jane Lusby (Oct 17 2020 at 02:10):

oh this is sick

view this post on Zulip Jane Lusby (Oct 17 2020 at 02:11):

I've been wanting to read about this kinda thing for a while

view this post on Zulip Jane Lusby (Oct 17 2020 at 02:11):

thanks for pointing it out @Jubilee

view this post on Zulip oliver (Oct 17 2020 at 02:19):

Oh if you're interested I usually don't like spamming theory papers but here is one
that just came across my desk and has some formalism around bidirection:
https://arxiv.org/abs/2010.07763

view this post on Zulip oliver (Oct 17 2020 at 02:25):

Seems alright to start I haven't read it all

view this post on Zulip Jane Lusby (Oct 19 2020 at 02:26):

definitely interested

view this post on Zulip Jubilee (Oct 25 2020 at 00:46):

my brain is roiling with thoughts about effect handlers enabling computational expressions a la F#.

view this post on Zulip oliver (Oct 25 2020 at 00:59):

isn't that what the backtrace does?

view this post on Zulip oliver (Oct 25 2020 at 01:00):

or rather the call to render the backtrace

view this post on Zulip Jubilee (Oct 25 2020 at 01:16):

possibly.

view this post on Zulip Jubilee (Oct 25 2020 at 01:20):

https://docs.microsoft.com/en-us/dotnet/fsharp/language-reference/computation-expressions the thing I'm thinking of is this, where basically you can define-your-own-syntax for "async" or "try" blocks.

view this post on Zulip oliver (Oct 25 2020 at 01:20):

the prototype we have now is write_backtrace_to(&mut dyn FormatterThing) -> Result<(),FormatterThing::Error>

view this post on Zulip oliver (Oct 25 2020 at 01:22):

there is talk about shimming (my word) in some try syntax


Last updated: Jan 26 2022 at 14:20 UTC