asyncRam' :: Enum addr => SClock wclk -> SClock rclk -> SNat n -> Signal' wclk (Maybe (addr, a)) -> Signal' rclk (addr -> a)

[strake]

and likewise asyncRam.

The path from read address to output is combinational, which we can so encode in the type.

My use case is a register file where we have a Signal (Either RegNum Word), either a register number or an immediate value, and we want a Signal Word, either the value in that register or the immediate value. It is much more convenient and less heinous with the proposed type of asyncRam:

liftA2 (flip either id) (asyncRam _ _) :: Signal (Either RegNum Word) -> Signal Word

versus

(\ rsi ->
 let rs = either id (pure undefined) <$> rsi
 in liftA2 (\ x -> either (pure x) id) (asyncRam _ _ rs) rsi) :: Signal (Either RegNum Word) -> Signal Word

[christiaanb]

I assume another varient of asyncRam will also work for you. I'd like to keep the current signature for people who are new to Clash/Haskell.

[comonoidial]

What about adding a special operation on signal functions, so that current simple asyncRam can be converted into the more flexible proposed variant.
It would be something alike to currying on signals that needs to be a primitive in the prelude:

curryS :: (Signal a -> Signal b) -> Signal (a ->b)

Then you could write the following (with an extra flip because of unfortunate argument ordering):

liftA2 (flip either id) (curryS $ flip (asyncRam _) _) :: Signal (Either RegNum Word) -> Signal Word

[christiaanb]

@comonoidial, i dont think is is possible to write that function.

curryS f = g :- gs
  where
     g  = ???
     gs = ???

[comonoidial]

Right this function can not exist, although the other way around is trivial. Even with other Signal representations it seems you can not enforce all the properties to allow for this function.
Does this mean we should provide signal of function returning variants for all primitives where it is feasible?

[strake]

Does this mean we should provide signal of function returning variants for all primitives where it is feasible?

I would prefer that myself. Essentially this means it returns time-varying combinational logic, i believe.

[strake]

@christiaanb: Could we have another term, say asyncRamFn or somewhat, and define asyncRam ... = (<*>) (asyncRamFn ...)?

[christiaanb]

@strake: yes, that'll work. Will you write a patch?

[strake]

Yep, once i figure out how to do so ☺