is rhsOf an effect or a query?

[andygill]

Translating

rhs-of 'binder

we can have two alternatives

rhsOf 'binder :: Effect () -- goto the right hand side of binder

or

rhsOf 'binder :: Effect LocalPath -- return the path to the right hand side of binder

The second one is cleaner, more modular, but more of a hassle to use; for example:

goto =<< rhsOf' binder

[xich]

In the current shell we overload it so it is both. That is, you can:

focus (rhs-of 'foo) (some rewrite)
rhs-of 'foo -- go to foo

It has a type like the second alternative you listed. Top-level queries
that return paths are just interpreted specially as 'goto's
On May 28, 2015 12:15 PM, "Andy Gill" [email protected] wrote:

Translating

rhs-of 'binder

we can have two alternatives

rhsOf 'binder :: Effect () -- goto the right hand side of binder

or

rhsOf 'binder :: Effect LocalPath -- return the path to the right hand side of binder

The second one is cleaner, more modular, but more of a hassle to use; for
example:

goto =<< rhsOf' binder

—
Reply to this email directly or view it on GitHub
https://github.com/ku-fpg/hermit-ghci/issues/1.[image: Web Bug from
https://github.com/notifications/beacon/AABNAGOjFgCOT8XLWVMgT1bu0QRXLyHwks5oNkhvgaJpZM4EsouY.gif]

[andygill]

So its not so much overloading (both rhs-of uses have the same type) but a property of the top-level shell. So we will need to support rhsOf :: Shell Path

But in the Black shell, you could want to examine the path, using

path <- send (rhsOf "foo") 

What is the semantics of rhs-of 'foo >> rhs-of 'bar. If its a top-level special effect, then you go to bar, otherwise you goto foo (perhaps passing a bar), then bar.

I think if we support the ::Shell Path we need to either

• Add a setPath/goto primitive; or
• Add a new type of send (again reusing goto).

goto :: Shell Path -> IO ()