James Bornholt

Yeah, we actually do set those timeouts, here: https://github.com/awslabs/mountpoint-s3/blob/b632bbe9645f1f6af26ed839e791b8a34ab74b36/mountpoint-s3/src/fs.rs#L201-L216 We set them very low because we want to [preserve S3's strong consistency model](https://github.com/awslabs/mountpoint-s3/blob/main/doc/SEMANTICS.md#consistency-and-concurrency) by default. But we know for some...

Absolutely!

Just to add that several of our other SDKs already send `Content-Length: 0` for CreateMultipartUpload (I tested boto3), so I'd expect adding this to work fine in terms of S3...

Yep, I think we're just waiting for https://github.com/awslabs/aws-c-auth/pull/237 and then we'll pull in all the CRT updates together so we can do just one new Mountpoint release.

Quick update on this one: we were planning to release this as part of 1.7.0 on Friday (https://github.com/awslabs/mountpoint-s3/pull/876) but we found an issue with a different change while testing the...

Yeah, Mountpoint requires additional permissions that Fargate isn't able to provide. For now, ECS + EC2 is our recommended option, but we're looking into what else might be possible here....

I kind of think we might prefer to keep the prefetcher returning `ChecksummedBytes`. Today we just convert immediately, but there's nothing stopping future us from inserting more work between the...

Ok, but now I'm not sure why we need this change — does moving the error parsing logic out into a `From` impl buy us anything? I reverted this after...

I looked at this a bit more carefully. It needs two more pieces of work on the CRT side: 1. The blocker right now is that the CRT's SSO provider...

This is cool, but `bincode2` hasn't been updated for 4 years and looks unmaintained, so not sure we should be switching to it.