You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At some point it would be nice to support optimisations. I think that some form of gradual typing / type inference would help here. For many scheme expressions the latent / dynamic type can be inferred at compile time. This would then allow for further optimsations and improvements.
Constant foldig
Direct calling to target methods avoiding the Func<object[], object> if argument types are known
Liveness tracking and escape analysis?
Some of this could be driven by the idea of 'type predicates' (integer?, vector?, etc. ) in a manner similar to TypeScript.
There is also discussion of 'assumptions': https://srfi.schemers.org/srfi-145/srfi-145.html that might be useful to enable more powerful optimisations in a given block.
The text was updated successfully, but these errors were encountered:
At some point it would be nice to support optimisations. I think that some form of gradual typing / type inference would help here. For many scheme expressions the latent / dynamic type can be inferred at compile time. This would then allow for further optimsations and improvements.
Func<object[], object>
if argument types are knownSome of this could be driven by the idea of 'type predicates' (
integer?
,vector?
, etc. ) in a manner similar to TypeScript.There is also discussion of 'assumptions': https://srfi.schemers.org/srfi-145/srfi-145.html that might be useful to enable more powerful optimisations in a given block.
The text was updated successfully, but these errors were encountered: