Remove bucket nameing rules restriction?

[seanses]

In an experiment effort to build an S3-compatible service I found it unnecessary to restrict bucket names according to https://docs.aws.amazon.com/AmazonS3/latest/userguide/bucketnamingrules.html. Does it make sense to add a quick path to disable check_bucket_name?

[Nugine]

Could you introduce your use cases of disabling check_bucket_name?
Note that S3Path type is a public enum which can be constructed from any string. We are not limiting its custom usage.

[seanses]

Thanks for replying! The GitHub notification is somewhat broken here..

I'm building an S3 protocol compatible endpoint for a general object storage which doesn't restrict bucket names according to the rules. The pain is that the request are rejected in prepare before handling by my service that impl S3.

My understanding is that the protocol doesn't put any restriction to the bucket name and the AWS implementation adds the rules. So my proposal is to defer any restriction checking to the implementations. What do you think?

[Nugine]

Sounds reasonable. I have two ideas:

• Add a specific option in S3ServiceBuilder for disabling check_bucket_name.
• Add a general validator mechanism for custom needs like this.

Which one do you prefer? Or is there any better design?

You can also try to implement in your fork in advance and see whether it fits your use cases.

[seanses]

I guess the first is the easiest to implement and that's what I'll need. I don't see where I would need to validate the bucket name (or prefix names). But anyway, I'll implement in my fork and try to make a PR. Thank you!

[Nugine]

Moved to issue #143