Figure out an alert and notification system

[fazo96]

No description provided.

[jbenet]

Pub sub will be very useful here.

We could probably bite the bullet and try a preliminary "very alpha"
version to try out for you.
On Tue, Nov 17, 2015 at 14:31 Enrico Fasoli [email protected]
wrote:

—
Reply to this email directly or view it on GitHub
#27.

[fazo96]

Yeah that could be veeery useful! Ihad to implement IPNS polling but on 0.3 it works very bad because the response times from go-ipfs are huge. I also have another problem currently which is user discovery, it works more or less but it's not good enough, also I need to resolve every single IPNS address of every possible user to get his data and that's very slow atm. I think it's probably going to be a lot better on 0.4 but I won't migrate my API until the files api is in place in js-ipfs-api

The issue at hand though is an interface to show notification (temporary or permanent). Temporary could be an error message or "settings saved" "post published" message, and permanent would be like github and facebook's notifications. I paused this because I wanted to dedicate the next few days to other more important issues

[jbenet]

yeah pub/sub will do it. I'll lift it up in my priorities.

On Thu, Nov 19, 2015 at 1:57 AM Enrico Fasoli [email protected]
wrote:

Yeah that could be veeery useful! Ihad to implement IPNS polling but on
0.3 it works very bad because the response times from go-ipfs are huge. I
also have another problem currently which is user discovery, it works more
or less but it's not good enough, also I need to resolve every single
IPNS address of every possible user to get his data and that's very slow
atm. I think it's probably going to be a lot better on 0.4 but I won't
migrate my API until the files api is in place in js-ipfs-api
ipfs-inactive/js-ipfs-http-client#74

—
Reply to this email directly or view it on GitHub
#27 (comment).

[fazo96]

The best thing would be the ability to listen to ipns update events, of every node. I'm scared about scalability of this approach though... what if there's a million nodes with hundreds of them updating every few seconds?