Too many fetches

[fauno]

I added console.log to signedFetch() and mentionToActor. When performing an approval action for a single activity, I see all of these requests logged.

https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
fetchhttps://skel.sutty.local/about.jsonld#main-key
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://skel.sutty.local/about.jsonld#main-key
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://skel.sutty.local/about.jsonld
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://fedi.sutty.nl/users/fauno
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://fedi.sutty.nl/users/fauno/inbox
https://skel.sutty.local/.well-known/webfinger?resource=acct:[email protected]
https://fedi.sutty.nl/users/fauno

[RangerMauve]

Yeah we really need a caching layer with a TTL.

[fauno]

Nice to have by the end of M4 if we can, but low priority.

[RangerMauve]

This could be used to replace the default fetch: https://www.npmjs.com/package/node-fetch-cache

We should use the storage location via the fs.

We should also have a large TTL, maybe a few hours at least? Even a day?

[fauno]

We should also have a large TTL, maybe a few hours at least? Even a day?

from what i've seen so far while developing caching on the side of the jekyll plugin, mastodon goes a good job at caching (it uses rails), so letting the caching library follow the remote instance headers for ttls should be ok