RFC for Fn::Length

[juegong2]

Language Enhancement Request For Comment

This is a request for comments about new intrinsic function Fn::Size. See #70 for additional details.

---

Licensing

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[jlhood]

@juegong2 Make sure to send a separate PR to update the README adding this RFC to the table.

[jlhood]

Great first draft. No issues with syntax. Just want to make sure we include some clarifications on limitations.

[jlhood]

2 minor comments, then it should be good to go!

[jlhood]

Looks good. Thanks for iterating!

[lorengordon]

I'd suggest Fn::Length for greater similarity with most languages I've seen. I had no idea what was meant by Size when I first saw it here. I thought maybe it was related to something like the "size" of an EC2 instance type.

[jlhood]

Can you put this change in a separate PR? If you couple it with this one it won't get added to the README until after it's fully approved which doesn't make a lot of sense, since it's in review right now.

[josb]

@lorengordon I'm fine with the identifier Fn::Length if that is preferred.

[jlhood]

Some data to support that "length" tends to be more popular than "size":

1. Java - Arrays have length property, Collections have size() method.
2. Javascript - Arrays have length property
3. Ruby - has both length and size methods (do the same thing, just aliases)
4. Python - has built-in len() function
5. Golang - also uses len() function
6. HCL - length()
7. CUE - len()

[juegong2]

Updated the naming of the new intrinsic function to Fn::Length

[jlhood]

Can you rename the RFC file as well?

[juegong2]

File name is updated as well

[benbridts]

I'm not sure if I understand the use case of this function:

• For conditionals it's very clunky without support for integer comparisons.
• For OutOfBound checking it has similar problems:
  • You can't use conditionals (see above)
  • CloudFormation will throw an error either way

[benbridts]

I would remove this note, or be very careful about wording. Promising a short-term solution and not meeting this self-imposed deadline is a good way to lose trust.

Additionally, getting this to be useful during provisioning looks complex from the outside:

Having !Length work during provisioning does seem similar to how other functions work (so short term might be okay here)
However, most use cases will expect to be able to use an !If with !Length, and this would be a major change in how CloudFormation works. Currently in the If documentation, there is this remark.

You can only reference other conditions and values from the Parameters and Mappings sections of a template. For example, you can reference a value from an input parameter, but you can't reference the logical ID of a resource in a condition.

Repeating that in the limitations here might be a good idea too

[juegong2]

Thanks for your feedback! I have updated the wording of this note.

[benbridts]

👍🏻