Fixes a bug where due to parallel file imports, the reported size of
ingested bytes was getting mixed up.

We now tally imported bytes against the reported file name instead of
assuming a sequential import, taking care to not let the tally grow
until memory is exhausted.

Aligns behaviour between js and go when no path is present during an import
- we now pass an empty string instead of `'unknown'`.

Replaces `SoL-arnet` bootstrap nodes from Node.js config to the new `dnsaddr` per #3370 (comment) and libp2p/js-libp2p-bootstrap#98

These new addresses resolve into DNS TXT records with tcp and wss based addresses for each peer (with `ipv4` and `ipv6`.

We claim to support URLs, MultiAddrs, strings and individual URL components
so add tests for them.

Also, add code to actually support the above permutations..

Co-authored-by: Vasco Santos <[email protected]>

Removes all unused dependencies, adds missing ones.

Also runs prod & unused dep checks in CI.

The browser fetch api doesn't allow reading of any data until the whole request has been sent which means progress events only fire after the upload is complete which rather defeats the purpose of reporting upload progress.

Here we switch to XHR for uploads with progress that does allow reading response data before the request is complete.

Co-authored-by: achingbrain <[email protected]>

If raw-leaves is unset, default to true if cid version is > 0.

This is in line with go-IPFS and is what the importer used to do internally.

 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]
 - [email protected]