Web API: sessionStorage

[aduh95]

Implements window.sessionStorage for Deno.

This is mostly a wrapper around Map, that should be 100% compatible with the browser implementation.

localStorage is a bit tricky to achieve on Deno, as far as I know their is no obvious equivalent to an origin on Deno, I have chosen to make it throw a SecurityError as the specs allows:

If the request violates a policy decision (e.g. if the user agent is configured to not allow the page to persist data), the user agent may throw a "SecurityError" DOMException instead of returning a Storage object

Missing features: window.localStorage, StorageEvent class and storage event.

Refs: https://html.spec.whatwg.org/multipage/webstorage.html#storage-2
Refs: #1657

[nayeemrmn]

-1 if this isn't actually persisting anything... certainly #1657 shouldn't be closed, and this is going to cause confusion as to whether or not this is properly implemented.

[kitsonk]

Yes, I say we need something that persists, even session storage. Until that point, any sort of "in memory" compatibility should be in deno.land/x.

[kitsonk]

Again, I don't think we should consider something that doesn't actually persist... maybe session storage can be in memory, but if all we have is in memory, we shouldn't add it at the moment, and keep something like this in deno.land/x/

We really need to work out the semantics of the origin for the storage... a lot of those are discussed in #1657 but not addressed here yet.

[kitsonk]

While technically, web workers don't have access to local storage, I would not want to add another bag here. We already have things in worker global that shouldn't technically be there. @bartlomieju opinions?

[bartlomieju]

I agree with you @kitsonk - workers shouldn't have access to localStorage

[aduh95]

I have updated the OP saying this PR doesn't actually implements localStorage. I think we should not disregard this PR just yet; although it's true it doesn't address the points discussed in #1657, my hope is that it can make the ball rolling for an actual implementation of persistant database in Deno.

[bartlomieju]

Thanks for opening the PR @aduh95 and sorry for slow review.

I agree with @nayeemrmn and @kitsonk opinions - this module should start as a 3rd party module in deno.land/x/ and make its way into public API after we figure out origin aspects. Closing without merge.