i don't know if there's a built-in means. i would just use a python script to programmatically pull
i don't know if there's a built-in means. i would just use a python script to programmatically pull the Images and push to R2
Bprobably a very naive approach 
ELooks like that is gonna be the way. But I read somewhere that Cloudflare Images is also using their R2 under the hood, so I am trying to confirm like if there is any other way than doing this manually.
Cif it's helpful, they're going to be changing the pricing of Images on November 15th and combining Images and Image Resizing: https://blog.cloudflare.com/merging-images-and-image-resizing/
I don't think that affects raw Images pricing though, If you are big enough, you could consider going Enterprise and talking to Sales
I don't think that affects raw Images pricing though, If you are big enough, you could consider going Enterprise and talking to Sales
The Cloudflare Blog
We’re changing how we bill for Image Resizing to let you calculate your monthly costs more accurately and reliably. All Image Resizing features will be available under Cloudflare Images
EHmm looks like new pricing will be more expensive if I use image resizing + r2. Did not know about this blog.
EIt is interesting that cloudflare has decided going with unit numbers for image count or resize count on this product. We will probably look for alternatives that relies storage size.
EAre you sure? The new pricing also makes major changes to how resizes are counted, its only unique transforms so if you resize the same image in the same way multiple times youre only billed once per month, rather than each time like currently.
EThe new pricing should be cheaper for almost every use-case
EI am just trying to imagine our use-case which is in a mobile app generally we show different small-medium-original size of images because of the performance, this is generally how it is in mobile apps. And we currently have over 3M+ image delivery per month including all variants of the image sizes.
EI see
EActually it would be much clear for us to calculate things if we were able to see currently on Cloudflare images the stats of what is the percentage of same image variant (size) out of total delivered image
SDo presigned urls work with the custom domains? I can't find any examples out there, and haven't been able to get it to work myself (401s).
HThey do not, no
SThat would explain it - thanks. Definitely feels like something that should be in the docs.
EGitHub
This keeps coming up in the Developer Community
KI have a question, what is the max. limit for resources that can be cached which are stored in R2, is it object level or entire bucket level
HThere isn't any limit, objects will just get ejected from cache at some point
HAs in, you are never guaranteed cache space(by number of objects anyway)
HSize of objects can determine whether it can be cached at all, but that is something else
YS3 pre signed URL, how to cross domains?
Bis it more efficient to HEAD a file that may not exist or fail to GET it?
Beg. would adding a HEAD to the beginning of reads cause a problem?
A
AHow do I get Class a Operations? in R2
ACan anyone help me ?
U
AHow to ?
UIt's at the bottom of the screenshot these are the class A operations:
ListBuckets, PutBucket, ListObjects, PutObject, CopyObject, CompleteMultipartUpload, CreateMultipartUpload, ListMultipartUploads, UploadPart, UploadPartCopy and PutBucketEncryption.
ListBuckets, PutBucket, ListObjects, PutObject, CopyObject, CompleteMultipartUpload, CreateMultipartUpload, ListMultipartUploads, UploadPart, UploadPartCopy and PutBucketEncryption.
ADoes that mean you automatically enter class A?
DA HEAD request counts the same as a GET request as far as both being a class B operation. So checking if an object exists before actually getting it will be 2 class B operations (and probably make things slower overall because you are making 2 API calls that aren't run in parallel).
ACan't buy class A straight away?
Bwas thinking that too
Bi do head before a put though
D"Class A" operations aren't a higher/faster tier you are trying to get into. Class A (generally) are write operations, class B are (generally) read operations.
Aahhh i see
Athanks
DYa, depends on your application. If you need to make sure you aren't overwriting something inadvertently, it would make sense to do a HEAD before a PUT.
Kdepending on what you're looking for in the
HEAD (i.e if it exists at all, or for modified date/etag) then you could use conditional uploadsBit's deterministic and overwriting isn't the end of the world but avoiding class A is the goal there
Bi actually keep a cached list of bucket contents and remove entries when they are deleted, so that i can avoid having to do a head or list_objects etc, for the most part. but i still have it there for the more expensive write operations, as they are operating in parallel
Bthere's more than 800,000 objects in these buckets
ᓚhi there! anyone knows where can I report an abuse commited in r2? Cloudflare team is not paying attention to my claims.
ᓚthanks in advance 
U?abuse
FIf you feel that a site is engaging in illegal or inappropriate activities, you can submit an abuse report at https://abuse.cloudflare.com/. The Trust and Safety team will then review the details and reply if appropriate. You can also report the site to your relevant local authorities. Reports cannot be filed via Discord or individual employees.
ᓚa phishing attack, i already reported through abuse.cloudflare.com with no success
UAs the abuse message says there is nothing we can do via discord. It has to be through the abuse form.
ᓚnoted, thanks
Pis there support for programatically creating API keys that are scoped to a specific prefix in a bucket? the use case is for allowing an untrusted user to upload a large file to a specific prefix. it seems that to use presigned URLs + multipart uploads, our API would have to create a URL for each part being uploaded which seems a bit messy
UI’m not sure if you need to create many urls for the multipart upload or not but if that is the case it’s just a client side operation so creating up to 10k parts which is the max should easily be doable in the initiate imo 
YWhy can't I submit data in the browser when I pre sign the URL? Continuously prompting for cross domain issues, how to solve them, or does R2 not support cross domain put?
