You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Right now we try window.ipfs and fallback to js-ipfs-api with the default config. If that fails to connect, we just retry until it does, without telling the user whats going on. This could be so much better.
When we're first run, if window.ipfs is available, we should let the user know that we're using it and that they are clearly wizards for having it installed. If not, we instantiate a js-ipfs-api (as we do now) on the default URL request the the node id with a short timeout. If we get no response, we should communicate with the user... we can ask if they just need to restart their local ipfs daemon, or provide a different API url if they're running with a customised config, or would they like to run an embedded js-ipfs node?
The "would you like to run a js-ipfs node" could be framed as "Hey, install ipfs-companion, and it'll run a js-ipfs node for you." which'd be great for ipfs-companion visibility... and could give us a bunch more users. On the flipside, it's out of band and adds a little friction when and we could just ask if they want us to install a js-ipfs node in a service worker at the click of a button. Either option would be rad. I'm leaning towards "hey why not install companion" as I think it's a big win to be visible in the browser.
The text was updated successfully, but these errors were encountered:
We need to remember that regular user won't know what service worker is, and I don't think we should mention js-ipfs as a viable option until it has the same connectivity magic (dht, relay discovery) as go-ipfs. Due to this installing Companion alone does not solve the problem of missing node yet (IMO).
A safe way to communicate things today would be to do basic heuristic to find IPFS API[1], and if that fails, prompt user to "manually start local node at port :5001 or install IPFS Desktop".
We can refine this after poc integrations with go-ipfs and companion land.
Right now we try
window.ipfs
and fallback to js-ipfs-api with the default config. If that fails to connect, we just retry until it does, without telling the user whats going on. This could be so much better.When we're first run, if
window.ipfs
is available, we should let the user know that we're using it and that they are clearly wizards for having it installed. If not, we instantiate a js-ipfs-api (as we do now) on the default URL request the the node id with a short timeout. If we get no response, we should communicate with the user... we can ask if they just need to restart their local ipfs daemon, or provide a different API url if they're running with a customised config, or would they like to run an embedded js-ipfs node?The "would you like to run a js-ipfs node" could be framed as "Hey, install ipfs-companion, and it'll run a js-ipfs node for you." which'd be great for ipfs-companion visibility... and could give us a bunch more users. On the flipside, it's out of band and adds a little friction when and we could just ask if they want us to install a js-ipfs node in a service worker at the click of a button. Either option would be rad. I'm leaning towards "hey why not install companion" as I think it's a big win to be visible in the browser.
The text was updated successfully, but these errors were encountered: