# clone
git clone https://github.com/guybedford/node.git --branch irp-implementation --single-branch node-esm
source |
---|
Contributors: Geoffrey Booth (@GeoffreyBooth), John-David Dalton (@jdalton), Jan Krems (@jkrems), Guy Bedford (@guybedford), Saleh Abdel Motaal (@SMotaal), Bradley Meck (@bmeck)
I have been trying to find a good way to paraphrase intuitive concerns regarding scheme-less specifiers and I think that while import:
is different from other ideas proposed for somewhat related purposes, it may be coming from similar but more tangible consideration.
Yesterday I did a little experiment with URL in node, Chrome, and Safari , and Firefox. My general thought is that new URL(x, y)
is a resolution formula that follows closest to the web standards so my assumption is that it pairs well with the whatwg expectations on specifiers.
The bottom line is that constructing scheme-less specifiers that depend on scheme-less referrers is not possible, and the hack used just to get through the list of valid specifiers was to resolve the referrer against file://y:/
which is not a proposed solution by all means.
Given the importance of specifier resolution (ie x -> y = z
) I am
module.exports = 'cjs'; |
Okay, concept means messy and probably requires flags on nightly or the likes, agreed? Just don't expect this to be more than what it is!
Live Demos
throttle — renders n instances of a perpetually changing element
https://smotaal.github.io/experimental/fragma/alpha/throttle#n
Along with the many disruptive trends in code editors and text documents (Markdown) in general (Markup) comes the inevitable question, Is it time for a new take on the various design patterns and implementation strategies used to work with source text?
Code Editors
While there are many JavaScript code editor implementations out there, you can almost divide them down by eras, for instance, ones predating GitHub's overwhelming effect on the adoption of Markdown, ones during the height of JQuery… etc. But regardless of eras, the way those editors model markup, while very much full of a lot of originality and creativity, they often seem to pivot around variants of similar and familiar patterns. One clear observation that begs pointing out, since most (if not all) value features in any implementation are often coupled to the specific variations of the markup model, adopting new and distruptive ideas often times leads to completely new implementations (some still call those ve
Option | Default | Description |
---|---|---|
validateOptions | false | crashes if mis-typed JSC options were passed to the VM |
dumpOptions | 0 | dumps JSC options (0 = None, 1 = Overridden only, 2 = All, 3 = Verbose) |
configFile | "" | file to configure JSC options and logging location |
useLLInt | true | allows the LLINT to be used if true |
useJIT | true | allows the executable pages to be allocated for JIT and thunks if true |
useBaselineJIT | true | allows the baseline JIT to be used if true |
useDFGJIT | true | allows the DFG JIT to be used if true |
useRegExpJIT | true | allows the RegExp JIT to be used if true |
This prototype simply relies on async execution to walk the module graph and uses the platform to create either a ModuleWrap in node (when possible) or a blob which is used to determine the actual specifier to use for each link outwards.
A given limitation is that circular references will result in deadlock.