AbstractBase case sensitivity about aws-sdk-ruby HOT 5 CLOSED

I have strong reservations against making AWS::Record case-insensitive. Both SimpleDB and DynamoDB are both case-sensitive. For this to work, the ids would have to be always normalized to lower-case strings. As support is added for custom id attributes with AWS::Record, this could cause a number of issues.

Have you considered filtering the incoming data from 3rd parties and normalizing the case as it is received from the source?

from aws-sdk-ruby.

I definitely understand your reservations, and the reasoning makes sense.

I guess my concern is that I am using HashModel, and I don't have easy control over the id. Is it guaranteed to always be lower case? I guess my concern is over this protected method, which I was planning on overriding in a subclass:

def populate_id
  @_id = UUIDTools::UUID.random_create.to_s
end

However, I don't like the idea of setting @_id in my subclass. What if we had a separate protected method, called generate_id, with the default implementation looks like this, with the populate_id changed appropriately:

protected 

def generate_id
  UUIDTools::UUID.random_create.to_s
end

def populate_id
  @_id = generate_id
end

This would allow me to override generate_id in my subclass, ensuring the same case all around?

Would you be amenable to this change?

from aws-sdk-ruby.

The plan was to make id a user-manageable attribute (this has not happened yet). The default randomly generated UUID is meant to simulate how a RDS can create an id, but without being monotonically increasing (this is generally a bad idea in no-sql systems).

The uuids are hexadecimal digits (0-9a-f), so it would be trivial to enforce that they are always lowercase (there is not plan to change this).

from aws-sdk-ruby.

I guess my question is this: does UUIDTools::UUID.random_create.to_s always return an all lowercase UUID? If not, I need to manage the id on my own (basically, calling @_id = UUIDTools::UUID.random_create.to_s.downcase in populate_id), which is not desirable. I can always just .downcase the incoming UUID from 3rd parties if the generated id is guaranteed to be always be downcased.

If [0-9a-f] is in fact the class of characters — and not [0-9a-fA-F] — then downcasing the id from the 3rd party will work. However, I need to be sure that is the case, either through documentation or controlling it on my end. Does this make sense?

from aws-sdk-ruby.

Yes, it will always return a lower case string (it will not return mixed case or upper case). Your concern does make sense. I'm going to go ahead and close this issue for now, as there doesn't appear to be anything that needs to happen in SDK code.

from aws-sdk-ruby.