It's pretty close to a KV Binding, docs aren't great right now though, pretty much only options are

It's pretty close to a KV Binding, docs aren't great right now though, pretty much only options are looking at existing code i.e cherryjimbo's sharex worker (https://github.com/Cherry/ShareX-R2-Cloudflare-Workers/blob/main/src/routes.mjs) and the Cf Worker Types which includes typings for all the r2 stuff

CChaika It's pretty close to a KV Binding, docs aren't great right now though, pretty mu...

Hello, I’m Allie!•4/30/22, 5:00 AM

You can also just use the dev branch docs: https://5141e50d.cloudflare-docs-7ou.pages.dev/r2/

Cloudflare R2 documentation · Cloudflare R2 docs

Store large amounts of unstructured data without egress fees.

ChaikaOP•4/30/22, 5:06 AM

Oops, there is some info on the docs. As far as I can see that's the same info as it on the live docs, and still is using outdated stuff like the internal bindings

I.e get mentions
const { value } = await MY_BUCKET.get(key);

Which doesn't work anymore

James•4/30/22, 5:22 AM

I’ve been meaning to make a PR to fix that one. That was the old internal bindings, yeah

Hello, I’m Allie!•4/30/22, 5:42 AM

R2 Docs(May be a little behind for now, while the team is constantly pushing changes):
https://developers.cloudflare.com/r2/

CCraiggles Can you guys pin this? Would be a good reference to add

Hello, I’m Allie!•4/30/22, 5:42 AM

Done

Erisa•4/30/22, 10:47 AM

repinned the first message to keep it on the top of the list

Ddarrennotfound Does r2 have any Api for usage info

John Spurlock•4/30/22, 1:06 PM

there are stats in the current analytics graphql endpoint: r2

John Spurlock•4/30/22, 2:04 PM

hehe

John Spurlock•4/30/22, 2:08 PM

i usually just send graphql queries constructed myself and process the response as any other json api, no generic framework needed for a known query or two

Ddarrennotfound But json needs ram right

Isaac McFadyen•4/30/22, 2:09 PM

Like a tiny amount.

Isaac McFadyen•4/30/22, 2:09 PM

There's no API that works without JSON, so that's your only option.

Ddarrennotfound Well it's still ram

Isaac McFadyen•4/30/22, 2:09 PM

Yeah but like maybe 2-4Mb?

Ddarrennotfound But graphql

Isaac McFadyen•4/30/22, 2:09 PM

GraphQL can be used with just regular JSON.

John Spurlock•4/30/22, 2:10 PM

eh its not so bad, you can summarize by time period to cut down on the data size you get back

Ddarrennotfound Isn't graphql a different thing than json

Isaac McFadyen•4/30/22, 2:10 PM

No.

Isaac McFadyen•4/30/22, 2:10 PM

GraphQL is just a format of JSON.

Isaac McFadyen•4/30/22, 2:11 PM

So you send a request, you get back json like

{
  "query": {...data}
}

{
  "query": {...data}
}

John Spurlock•4/30/22, 2:11 PM

worst case you could build a worker to do it : ) but an rpi can process json just fine in my experience

Ddarrennotfound Wait a minute But it does return lots of unsessary field

Isaac McFadyen•4/30/22, 2:11 PM

No it doesn't.

Isaac McFadyen•4/30/22, 2:11 PM

With GraphQL you can filter the fields you get back.

Isaac McFadyen•4/30/22, 2:11 PM

That's the point of GraphQL.

Isaac McFadyen•4/30/22, 2:11 PM

Also @darrennotfound we should move to #off-topic if we keep discussing this.

John Spurlock•4/30/22, 3:36 PM

this morning I tried getting litestream (sqlite replication) working with r2

John Spurlock•4/30/22, 3:36 PM

good news and bad news

John Spurlock•4/30/22, 3:37 PM

good news is that, it currently works!

John Spurlock•4/30/22, 3:37 PM

bad news is that it requires ListObjectsListObjects v1, which is not supposed to be implemented right now in R2, and is only kind of working due to a bug : (

John Spurlock•4/30/22, 3:38 PM

added details on litestream config here: https://gist.github.com/johnspurlock-skymethod/6027c81953f81aa535d889d86a1061ba#litestream-sqlite-replication

John Spurlock•4/30/22, 3:39 PM

at this point i would consider proper support for ListObjectsListObjects v1 to be a high priority fix before any wider r2 release

beelzabub•4/30/22, 4:34 PM

Is the main difference that start-after is called marker and there’s no continuation token?

beelzabub•4/30/22, 4:59 PM

Or I guess marker is a mix of both

John Spurlock•4/30/22, 5:14 PM

i think start-after and marker are semantically equivalent, no? big change is continuation-token, also owners should not be included in v2 results by default, and can be re-enabled with fetch-ownerfetch-owner

beelzabub•4/30/22, 5:38 PM

I generally struggle to see how continuation-tokencontinuation-token is really different from markermarker aside from the fact that it's an opaque value (i.e. how would a hypothetical NextStartAfterNextStartAfter response value that you feed into start-afterstart-after be different from NextContinuationTokenNextContinuationToken).

beelzabub•4/30/22, 5:39 PM

i.e. I'm intellectually curious about why they split it up. The fetch-ownerfetch-owner thing I get since egress bytes are charged on S3 (but no one is complaining about the use of XML instead of a binary protocol or even JSON?)

Erisa•4/30/22, 5:56 PM

oh there was a reply here https://github.com/rclone/rclone/issues/5422#issuecomment-1113971154

GitHub

Include X-Amz-Content-Sha256 header for all requests to S3 (current...

The associated forum post URL from https://forum.rclone.org I asked on the forum, but I haven't gotten any comments there yet unfortunately https://forum.rclone.org/t/send-x-amz-content-sha...

Erisa•4/30/22, 5:56 PM

i should subscribe to that issue