isiahmeadows at gmail.com
Fri Apr 5 08:33:24 UTC 2019
Yeah, I agree that until the ECMAScript spec defines any I/O APIs, it
doesn't belong in the spec. The spec is great for things that involve
pure data processing, but anything beyond that should be left to
embedders ("hosts" in the spec) to define.
I do feel that a separate spec dependent on ES should exist for stuff
that goes above and beyond simple data processing but is still broadly
useful in standard mobile/desktop/browser/server runtimes, like:
- Web requests, specifically just the `fetch(string, opts?)` part +
associated response APIs
- Native cryptography APIs
- WebAssembly integration
- Off-screen GPU access with permitted CPU fallback, useful for image
processing and large-scale data processing on server-side and
This would be things the *host* would implement, ideally as standard
library modules, not the *engine*. Each of these would be optional,
but the HTML spec should require hosts to implement most modules
to elaborate on this further.
contact at isiahmeadows.com
On Thu, Apr 4, 2019 at 9:18 PM Logan Smyth <loganfsmyth at gmail.com> wrote:
> Symbols seems like a good way to do this, but since the ECMA spec doesn't define anything about IO, I don't think this would be their responsibility. This seems more like something the console API spec would expose, e.g. `console.inspect`, where the `Symbol.XX` namespace would be reserved for ECMAScript-builtin symbols.
> On Thu, Apr 4, 2019 at 5:44 PM Sultan <thysultan at gmail.com> wrote:
>> Like Symbol.iterator, a Symbol.inspect symbol for use in implementing cross-platform console display introspection.
>> Currently node has something akin to this with a magic inspect method on objects.
>> This would pave a cow path for how different platforms can afford this ability to consumers without each inventing their own heuristic, i.e in the browser i might have an exotic object that i want console.error to display a toString payload instead of the objects shape.
>> es-discuss mailing list
>> es-discuss at mozilla.org
> es-discuss mailing list
> es-discuss at mozilla.org
More information about the es-discuss