yeah your right its not even in draft, i think in my opinion the benefit is you dont need to type/copy or write long url and you can easily find what are your dependencies. :)
another thing is for example (module 1 dependencies of module 2, but the module 1 didnt use the URL)
module 1
Pretty cool - this basically is the web equivalent of package.json 🙂
The main benefit of this is not about long or short URLs - it's about ensuring that, when several modules depend on the same things, they're also loading the same versions of those things. CDNs such as Snowpack try to address this by allowing URLs with version constraints, and performing redirects, but that's a half measure - it doesn't perform or scale well, doesn't let you lock to a specific version, etc...
We really need import-maps to proliferate into a standard. By then, for the first time in the history of the web, the platform will actually be able to stand alone, without all the clunky build tools and bundlers. 😄✊
That's what I'm hoping! Much of the Node.js ecosystem is designed to accommodate the missing features in JavaScript and the web browser. It's too bad, because while the transition is needed, it will be a long one.
I didn't even know that import maps were used in Chrome and Edge. That just leaves Firefox and Safari if browser developers wanted to implement this without a standard, though I would prefer the standard existed.
You want deterministic dependencies. More than once, a buggy update to a single, one-liner package has broken millions of projects in a day. It's a good thing someone else agrees.
Maybe in 5 years, all the features we're talking about today will replace NPM dependence entirely.
yeah your right its not even in draft, i think in my opinion the benefit is you dont need to type/copy or write long url and you can easily find what are your dependencies. :)
another thing is for example (module 1 dependencies of module 2, but the module 1 didnt use the URL)
module 1
module 2
I had thought that this would essentially replace the package.json file.
If so, then import maps bring us one step closer to building data-driven web applications without Node or NPM.
It's still not on standards track, it seems - though it is available and enabled by default in Chrome and Edge.
caniuse.com/import-maps
Pretty cool - this basically is the web equivalent of package.json 🙂
The main benefit of this is not about long or short URLs - it's about ensuring that, when several modules depend on the same things, they're also loading the same versions of those things. CDNs such as Snowpack try to address this by allowing URLs with version constraints, and performing redirects, but that's a half measure - it doesn't perform or scale well, doesn't let you lock to a specific version, etc...
We really need import-maps to proliferate into a standard. By then, for the first time in the history of the web, the platform will actually be able to stand alone, without all the clunky build tools and bundlers. 😄✊
That's what I'm hoping! Much of the Node.js ecosystem is designed to accommodate the missing features in JavaScript and the web browser. It's too bad, because while the transition is needed, it will be a long one.
I didn't even know that import maps were used in Chrome and Edge. That just leaves Firefox and Safari if browser developers wanted to implement this without a standard, though I would prefer the standard existed.
You want deterministic dependencies. More than once, a buggy update to a single, one-liner package has broken millions of projects in a day. It's a good thing someone else agrees.
Maybe in 5 years, all the features we're talking about today will replace NPM dependence entirely.
That's actually impressive that a feature from backbone and request.js is finally make it into webkit based browsers.