Re: how to extend a capability on a data source?
Rob Withers <[email protected]>
| Newsgroups | gmane.comp.capabilities.general |
|---|---|
| Message-ID | <[email protected]> |
On Nov 4, 2013, at 8:15 AM, David Barbour <[email protected]> wrote: > > On Mon, Nov 4, 2013 at 7:55 AM, Rob Withers <[email protected]> wrote: > > Why do I want to statically enforce some type in a dynamic, distributed, heterogeneous environment? Just dynamically enforce it and go to the next item on your task list. > > I can't say why you might want to. But the reasons I favor are: elegance and performance. If I can make a few useful guarantees at compile-time, I can avoid much unnecessary logic (and bugs) in the runtime, and I can avoid complex partial-failure handling code (and bugs) in the user's logic. Further, I can more easily compile for heterogeneous environments where dynamic enforcement is difficult or super expensive: FPGAs, OpenCL, shaders. > > Dynamic systems can be modeled as transitions from one well-typed program to another, at well-defined boundaries, with a controlled amount of slack for the input and output types. Dave, I must admit I like type inference, as well as the performance benefit from typing at compile time. Compare the performance between groovy and Scala. Is there any type inferred system which could type infer a promise as having the type of the return value of the function which was eventually sent to? The other issue, on the JVM anyways, is that a wrapper is required, since there are no mutable refs. thanks, rob _______________________________________________ cap-talk mailing list [email protected] http://www.eros-os.org/mailman/listinfo/cap-talk