`do` monad construct / `for` comprehension

[Kamirus]

Currently we have the Option block syntax do ? { ... } to avoid pattern-matches when dealing with multiple ?T values. Currently we are lacking a similar mechanism for other types, e.g. Result.

As an extensible solution we might consider For Comprehensions and For Loops from Scala, which is a syntax sugar that translates expressions like

for {
  x <- e1;
  y <- e2
} yield f(x,y)

Into a sequence of flatMap, map, filter, forEach etc.

We already have a for-loop syntax which would synergize with this extension. For-comprehension syntax is similar to the do monad syntax, but instead of relying on the Monad type class (which we don't have ATM) it simply requires the presence of a standard operation like flatMap/map etc.

---

Links to related issues and resources:

• WhitePaper.md

[ggreif]

We already had a few shots at this, there should be some PRs/issues around...

[Kamirus]

We already had a few shots at this, there should be some PRs/issues around...

@ggreif I just saw the meta-issue and a mention in the WhitePaper.md -- they are both linked in this issue. If anyone finds more references related to this topic please edit and add them or link via github.

[ggreif]

My preferred syntax would be

do with <module> { … };

and the monadic bind being let! val = <monadic term>;

But as @christoph-dfinity points out we already have special "monads"

• option do ? { ... };
• async/await

He would prefer to have a performant (i.e. short-cutting like do ?) version of the Result monad.