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
Proposal: consider url parsing to be more flexible about colons in the protocol part of a url enabled Dynamic -extension:// protocol resolved via serviceWorker
we can also name this:
Component based url specifiers
none network related extension specifiers
dynamic extension url handling extension://
dynamic url registration
Update
as i have writen tests for this i found out that we get near good results
// works
protocol: "node:"
// fails
protocol: "git+ssh@github:"
so its about allowing custom protocol specifiers with some additional signs
the most essential part of this is that a url that does not conforms to the networkProtocolName:// pattern gets parsed as is into
protocol and path propertys of the URL object as is without handling search parameters hash domain or anything
newURL("[email protected]:/me"){protocol: '[email protected]',path: '/me'}<protocolendswith : >:<pathHashSearch>/*for compat if path starts with // and protocol is http<s> we strip it from path and take the domain and port as done before but if the protocol is unknown we strip into protocol path and path it to the serviceWorker or what ever implements the fetch load of the ECMAScript tc39 Spec*/
fetch should use the ECMAScript resolve and loader implementations
so this is related to URL Object in tc39 and whatwg fetch and maybe whatwg url definition.
Why?
Browsers get more modular and even build custom for embedded devices this is at present complicated so the WinterCG got created to address needs of Web Interoperable Runtimes the most famouse of that runtimes is maybe Chromium or Webkit depends on the view. Both projects get now split into internal IPC Connected Components. it solves as referenced in the issue the problem of assertions that are not based on a file extension or a additional meta like
importmefrom'[email protected]:whatwg/fetch'// and notimportmefrom'whatwg/fetch'assert{type: "git+ssh@github"};
so it would be url serializeable
this is used in bundlers and runtimes already
importfsfrom'node:fs'
i want to turn url into a specifier resolve algo for better interop on the long run with other browser external or internal components as we overall design browser more modular today
now as a coder and web author it would be beneficial if he could define ways to interact with them directly from the dom in a declarativ way without coding additional web components. As also existing interop between url specifiers in NodeJS and Browsers is not given as new URL does not handle none http related urls well.
use cases
Accessing Components
fetch needs to be enabled for example to accept raw graphics pipeline input which it is able to do at present i am passing a memory fileDescriptor to it and stream the body but at present i bound that in a hardcoded way and all other components do so
it would be beneficial if we could fetch raw body streams and post to them directly via a unified url protocol
this would even enable circumventing the ECMAScript context. internal
this also enables a whole set of stuff that we transpil at present before the run
<!-- Current snapshot of the webgpu rendering result at time the element got inserted into the dom --><!-- directly copyed internal no round trips direct mem copy --><imgsrc="webgpu:0">
It will supply the fundamentals for linking the components cake layer into the dom via url specifiers
think about it as a dynamic unified version of the extensions:// protocol implemented in browsers
WinterOP
does enable again interop without transpilation with other ECMAScript Runtimes that are Web Inter Operable
so what i vote for is that we get it some how into any standard that is Winterop related that fetch should accept more then one : while taking only the last as protocol specifier
as i think more about all the possible combinations with ports this needs some refinement it is not the only rule to take the last ":// " maybe some more rules are needed like if there are more then one double point. Bit i guess you all get why we need that
custom protocol specifiers are one of the most best interop features today to get the browser and other runtimes into diffrent handling.
the only alternate to custom protocol specifiers are searchParms and URL Hash which are by fare less readable then protocol specifiers like
Enable Direct referencing browser components via url specifers
moved from:
Proposal: consider url parsing to be more flexible about colons in the protocol part of a url enabled Dynamic -extension:// protocol resolved via serviceWorker
we can also name this:
Update
as i have writen tests for this i found out that we get near good results
so its about allowing custom protocol specifiers with some additional signs
the most essential part of this is that a url that does not conforms to the networkProtocolName:// pattern gets parsed as is into
protocol and path propertys of the URL object as is without handling search parameters hash domain or anything
Proposal
Currently Fails
Expected
fetch should use the ECMAScript resolve and loader implementations
so this is related to URL Object in tc39 and whatwg fetch and maybe whatwg url definition.
Why?
Browsers get more modular and even build custom for embedded devices this is at present complicated so the WinterCG got created to address needs of Web Interoperable Runtimes the most famouse of that runtimes is maybe Chromium or Webkit depends on the view. Both projects get now split into internal IPC Connected Components. it solves as referenced in the issue the problem of assertions that are not based on a file extension or a additional meta like
so it would be url serializeable
this is used in bundlers and runtimes already
i want to turn url into a specifier resolve algo for better interop on the long run with other browser external or internal components as we overall design browser more modular today
now as a coder and web author it would be beneficial if he could define ways to interact with them directly from the dom in a declarativ way without coding additional web components. As also existing interop between url specifiers in NodeJS and Browsers is not given as new URL does not handle none http related urls well.
use cases
Accessing Components
fetch needs to be enabled for example to accept raw graphics pipeline input which it is able to do at present i am passing a memory fileDescriptor to it and stream the body but at present i bound that in a hardcoded way and all other components do so
it would be beneficial if we could fetch raw body streams and post to them directly via a unified url protocol
this would even enable circumventing the ECMAScript context. internal
this also enables a whole set of stuff that we transpil at present before the run
WinterOP
does enable again interop without transpilation with other ECMAScript Runtimes that are Web Inter Operable
see:
see:
npm ci
nodejs/node#44998 (comment)so what i vote for is that we get it some how into any standard that is Winterop related that fetch should accept more then one : while taking only the last as protocol specifier
as i think more about all the possible combinations with ports this needs some refinement it is not the only rule to take the last ":// " maybe some more rules are needed like if there are more then one double point. Bit i guess you all get why we need that
custom protocol specifiers are one of the most best interop features today to get the browser and other runtimes into diffrent handling.
the only alternate to custom protocol specifiers are searchParms and URL Hash which are by fare less readable then protocol specifiers like
vs
Allows implementation of custom protocol handlers
This enables dynamic protocol handler registration on the serviceWorker level which resolves the fetch requests.
The text was updated successfully, but these errors were encountered: