Not "I hope it's not deprecated"

I wonder how far along

aws-sdk-rs

is - I only really think about it for S3 (because of R2)

To add onto what @kian said above: I'd strongly recommend against doing SigV4 manually if you can at all help it, it's painful.

aws-sigv4

has a

from(request: &'a Request<B>) -> SignableRequest<'a>

so it's a lot like

aws4fetch

I suppose

messing around with rust-s3 now

hey, that worked!

ty folks!

one step closer to a functioning personal private crate registry

Ooo, that's cool

now i just need to clean up the code before I open-source it

I've never looked into what APIs that a crate registry exposes, I think a few people have done Docker registries on R2

It's pretty simple actually. If you ignore the default git registry and only support sparse, it's not hard to write one

Does

cargo

checksum the tarball locally?

R2 has some funky behaviour with large files (aka multipart upload) and how their checksums are stored, but I imagine if it's checksum'd locally then that isn't an issue

it verifies it on pull, you need to generate a checksum server-side on upload

and store it in the index

which in my case is a postgres db

Sounds like it'll work pretty nicely

Next step is using #882757739019911219 and #992060581832032316 instead /s

the other nice thing is that sparse goes well with cdn

> The sparse protocol downloads each index file using an individual HTTP request. Since this results in a large number of small HTTP requests, performance is significantly improved with a server that supports pipelining and HTTP/2.

Oh yeah, that sounds like it'll be cache's best friend.

I don't think I've tried sparse yet but I have stared at the updating index bar every now and then so I probably should

yea

sparse is especially nice in ci

I'm a bit worried about the billing - I can't see any charges or anything and when I go to billing there's no estimated prices or anything does that mean i've not used enough R2 to bill?

Hey guys, I just switched from Wasabi to R2 and I've been seeing a significant drop on my cache hit rate. It used to hover around 80% with tiered caching on but now with R2, it seems to be around 20%. Looks like assets are getting evicted faster than when I was using other storage. I have page rule set to cache everything and it was working fine before I switch to R2. Any one seeing this?

FYI, I'm using a custom domain with R2 so Cloudflare cache should work as it was before.

I'm trying to delete a bucket, but I need to empty it first. I´m inferring from https://discord.com/channels/595317990191398933/940663374377783388/1064259479857807390 that the --recursive s3 rm api it isn't implemented? The command below does not seem to work in R2, it works in S3, tho.

aws s3 rm s3://bucket_name/ --endpoint-url https://xxx.r2.cloudflarestorage.com/bucket_name --recursive --dryrun