A guide to IPFS connectivity in web browsers | Hacker News Reader

A guide to IPFS connectivity in web browsers

blog.ipfs.io

136 pointstosh30 comments

This is nice but for more mass adoption we'll need something seamless that will just be included as a script in a web page and "just work" by connecting to a provider while still customizable for people who want to use their own node.

It can even "polyfill" any IPFS paths automatically (which should just be simply as replacing any ipfs:// URIs with ipfs.io gateway paths).

For anyone trying to host pinned files behind a NAT/CGNAT would it be possible to hole punch to serve paths directly? If possible everything can me set up to work just out of the box with minimal/no setup in browser which should be the aim.

In other words, we need something of an IPFS version of what MetaMask/ethers.js did to Ethereum based blockchains.

> This is nice but for more mass adoption we'll need something seamless that will just be included as a script in a web page and "just work" by connecting to a provider while still customizable for people who want to use their own node. It can even "polyfill" any IPFS paths automatically (which should just be simply as replacing any ipfs:// URIs with ipfs.io gateway paths).

That’s what Brave does built in, and other browsers with the ipfs companion extension

When you hit a ipfs URL it’s resolving it either through a gateway or through your own node

That's great but the problem is still Brave's market adoption.

We'd really need some support from Firefox/Chrome/Safari trio to really start adoption. Most likely Firefox, then (albeit unlikely) Chrome and Safari.

Brave's adoption is just too low, even though I (and probably a significant people here ar HN) personally think it's doing the best.

Look at the trends. Brave has barely 3 years since it became usable enough to be a daily driver, and it has already 20-25% of Firefox market share. Firefox is constantly shrinking its user base, Brave is growing 40% YoY.

Where are you finding those numbers?

Most global browser share numbers I could find bundled Brave into the "Other" category - this was the only one I found that broke it out separately: https://kinsta.com/browser-market-share/

- Firefox at 216M MAU: https://data.firefox.com/dashboard/user-activity

- Brave at 54M MAU: https://basicattentiontoken.org/growth/

Currently working on Android + iOS apps that handle the ipfs:// protocol natively and allow users to upload files as well. I think you're right on target RE: more protocol adoption and hopefully something as simple as making the links work natively and letting people upload with a simple interface will help out. The IPFS extension exists for Chrome/Firefox/Opera for similar functionality.

Out of my hands but getting more browsers than Brave to add support seems like the next place to start pushing IMO.

The current state of the browser its just not possible to do p2p properly, WebRTC is the best bet and you need to trust stun servers to punch NAT you need to trust a signal server to connect to other peers its not real p2p so why not just trust a gateway and use that.

Brave browser has ipfs support built in and I think there are plugins for chrome and firefox without the code localy to do p2p properly it doesn't really make much sense faking it with WebRTC when the end result is the same as using a gateway directly.

Trying local node, verifying it works, otherwise falling back to an external node might be the best way to go. But still we need Firefox/Chrome/Safari adoption.

Do I understand correctly that I still need to run either an extra service outside the browser on my client machine, or some service on a central server I own (the WebRTC-Star thing?) for this to work?

For browsers that don't natively support IPFS yes.

It's worth noting that native support of IPFS is usually going to mean something like what Brave now does: the browser manages an instance of an ipfs-go node.

Will browsers that natively support IPFS automatically become distribution nodes? Or is cached-but-not-pinned data not served to other peers?

Is there any suggestion that browsers beyond Brave will support it?

Is a first-party IPFS Web Extension a possibility? Like how it would have been with the old-style Firefox extensions?

It's not a standard yet, but you can try it out in Firefox Nightly. https://github.com/mozilla/libdweb/blob/master/Readme.md#pro...

These libdweb Browser Extension APIs never landed in any official Firefox Nightly build. Mozilla stopped working on libdweb in 2019 (https://github.com/ipfs/ipfs-companion/pull/533#issuecomment...).

Ah dang. I figured it was popular enough that someone would be working on it at least experimentally.

Seems like its necessary to get the p2p stuff working properly other wise it all just uses a gateway.

https://addons.mozilla.org/en-US/firefox/addon/ipfs-companio...

What about upnp as an alternative to stun to achieve real p2p? You may need something native to setup but would not rely on a relay.

UPnP does not work on all networks (eg. double-NAT), and network administrators may be afraid to allow UPnP traffic as it would easily open doors to the LAN from the outside (for example from a stupid "smart" device).

> First step involves the command line

This is the "web of tomorrow"?

Well, yes - the key word in that statement being "tomorrow".

Technological innovation doesn't happen overnight. Browsers don't get IPFS support overnight. User interfaces don't get created overnight. And so, yes, sometimes it is up to people who can use the command line to pave the way to, as you're calling it, the web of tomorrow.

> This is the "web of tomorrow"?

I'd argue that those who never wanted to do any of these things pretty much got the "web of today" they deserve.

Don't worry the second step is to replace your keyboard with a neural link.

2022 is the year of <s>Linux</s>IPFS on the desktop!

The truly ironic thing about this comment is that there are more devices is users hands today that run Linux than run any other operating system. Of the remaining devices, the majority run a variety of BSD Unix.

All of these devices might not use a WIMP paradigm—opting instead for a touch interface—but for most users, the majority of the GUI interactions they have with computer applications are facilitated by a Linux or Unix operating system.

Linux won in the end, it’s desktop computing that lost.

Maybe something similar will happen with IPFS and related technologies?

Hate it when articles don't expand on abbreviations. Yes, I want to read this. No, I have no heckling what IPFS is.

It's an article about IPFS, hosted at a website also called IPFS.io. Just click the logo, and it will answer all of your questions.

https://docs.ipfs.io/concepts/what-is-ipfs/#what-is-ipfs

> That last point is actually where IPFS gets its full name: the InterPlanetary File System. We're striving to build a system that works across places as disconnected or as far apart as planets. While that's an idealistic goal, it keeps us working and thinking hard, and almost everything we create in pursuit of that goal is also useful here at home.

Reply on news.ycombinator.com