[w3c/IndexedDB] Site-agnostic databases based on `registerProtocolHandler` (#264)

I am wondering if, in place of the perhaps intimidating suggestion brought up through #75, an easier bridge forward to site-agnostic databases might be to tie the privilege to `registerProtocolHandler`.

A special format, such as extending (or replacing) the `web+` prefix, e.g., `web+indexeddb+`, might signify a protocol handler which also would grant access to a shared database as well.

So a site might request permissions like this:

```js
navigator.registerProtocolHandler(
    "web+indexeddb+com_example_dbformats_cdcollection",
    "https://cds.example.com/?collection=%s",
    "CD collection browser"
);
```

The variable `%s` need not be present, but it might be used to allow such
as flagging a default store or view of the collection for the site to open, e.g., 
`web+indexeddb+com_example_dbformats_cdcollection:folk-genre`.

The user would then be asked whether they wished to add `cds.example.com` as a handler for `web+indexeddb+com_example_dbformats_cdcollection` links and also be able to create or access a shared `com_example_dbformats_cdcollection` database (potentially informing the user of any other sites which had already registered for this database format).

If the user approves the protocol, the site could then begin using a shared database (and create the database if the site--or another site--has not already created one) by:

```js
const req = indexedDB.open('com_example_dbformats_cdcollection', undefined, {
  shared: true
});
```

(If the protocol access had *not* yet been created, this might throw or be a no-op.)

The existing UI for storing approved protocols could thus be leveraged.

If there is concern that a user might already have approved such a protocol before knowing it can now support shared database creation/access, the protocol name, e.g., `web+indexeddb+` might have reserved characters would not be permissible in the current scheme.

While there are of course different uses for web protocols and shared databases, I think they may be considered similar enough in both handling custom formats to justify the latter leveraging the former.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/IndexedDB/issues/264