nil inputs in service calls

[siggimoo]

I just happened to notice while reviewing the source code that passing a nil input to service calls seems to be allowed, with a zero-value struct being used in its place. For instance, the following two statements are effectively identical:

output, err := s3Service.ListBuckets(ctx, nil)
output, err := s3Service.ListBuckets(ctx, &s3.ListBucketsInput{})

While this is handy, I can't seem to find the behavior documented anywhere and I don't want to rely on it if it's not a standard convention. Does anyone know anything about this?

[RanVaknin]

Hi @siggimoo ,

I'm not sure what exactly your question is.
We allow nil inputs because the request code handles nil.

func (c *S3) ListBucketsRequest(input *ListBucketsInput) (req *request.Request, output *ListBucketsOutput) {
	op := &request.Operation{
		Name:       opListBuckets,
		HTTPMethod: "GET",
		HTTPPath:   "/",
	}

// You can see here that nil inputs get converted to an empty struct.
	if input == nil {
		input = &ListBucketsInput{}
	}

	output = &ListBucketsOutput{}
	req = c.newRequest(op, input, output)
	return
}

Thanks,
Ran~

[siggimoo]

I am aware that at least some of the code guards against nil inputs -- I haven't checked every single service call. What I'm not readily able to find is any mention of this in the documentation (e.g. here). So it's unclear whether this is behavior I can trust to always be there.

[RanVaknin]

Yes I'd say that it would be the case for 99% of the times since the SDK clients are auto-generated aside from a handful of hand-written ones.