(Moving this to a separate thread because it's not documentation-related and the original thread was about docs.)
The lack of public nodes is a great example of what I described in my previous post about priorities and how they affect devs.
So, Farcaster does not offer public API endpoints at the moment. It has practically outsourced this to Neynar. Which is great (we don't want a single entity to do everything, do we?).
But Neynar had to obviously set some limits, so they have introduced some minor changes in the requests (an extra header with an api key, and limit to response size), and the infra (the node sits behind a proxy of some sort that checks the api key.
The side-effect is your standard code, implemented according to the docs, that used to work with hubs does not work with Neynar's hubs, not without modifications.
Had the core FC team had to deal with the same problem, they would have probably ended up with the same solution Neynar did. But the solution would have been part of the API spec. And the code I wrote (take fcp for example) and tested with my local hub would work with Neynar out of the box.
Now,
a) I've spent hours trying to understand how to implement and debug the Neynar changes.
b) I can't tell a user, here is a program to back up your casts, get a free key from Neynar to use/test it.
c) Why even mention fcp, if I have to respond to most of the requests, "well, no you can't use it, unless you set up your hub, dedicate 2TB of disk space and wait for a couple of days to sync". 2 replies
1 recast
5 reactions
3 replies
0 recast
1 reaction
1 reply
0 recast
1 reaction
2 replies
0 recast
0 reaction
1 reply
0 recast
0 reaction
0 reply
0 recast
3 reactions
