CloudSearch is forced as a "GET"

[martinjt]

This means that large queries fail due to the "RequestEntityTooLarge" error.

Is there a way to allow the caller to change this? maybe using the ClientConfig?

[PavelSafronov]

Thanks for opening this issue. This is a known limitation with the GET method. At the moment, there is no way to configure the client to use POST instead, but we are adding this feature to our internal backlog and will keep this issue updated.

[martinjt]

Ok, well I worked around it by creating my own derived class and marshaller.

    public class CloudSearchClient : AmazonCloudSearchDomainClient
    {
        public CloudSearchClient(AWSCredentials creds, AmazonCloudSearchDomainConfig config) :
            base(creds, config)
        {
        }

        public new SearchResponse Search(SearchRequest request)
        {
            var marshaller = new PostSearchRequestMarshaller();
            var unmarshaller = SearchResponseUnmarshaller.Instance;

            return Invoke<SearchRequest, SearchResponse>(request, marshaller, unmarshaller);
        }

        public new Task<SearchResponse> SearchAsync(SearchRequest request, CancellationToken cancellationToken = new CancellationToken())
        {
            var marshaller = new PostSearchRequestMarshaller();
            var unmarshaller = SearchResponseUnmarshaller.Instance;

            return InvokeAsync<SearchRequest, SearchResponse>(request, marshaller, unmarshaller, cancellationToken);
        }
    }

and the Marshaller

    public class PostSearchRequestMarshaller : IMarshaller<IRequest, SearchRequest>,
        IMarshaller<IRequest, AmazonWebServiceRequest>
    {
        /// <summary>
        /// Marshaller the request object to the HTTP request.
        /// </summary>  
        /// <param name="input"></param>
        /// <returns></returns>
        public IRequest Marshall(AmazonWebServiceRequest input)
        {
            return Marshall((SearchRequest) input);
        }

        public IRequest Marshall(SearchRequest publicRequest)
        {
            IRequest request = new DefaultRequest(publicRequest, "Amazon.CloudSearchDomain");
            request.HttpMethod = "POST";

            string uriResourcePath = "/2013-01-01/search";
            request.Parameters.Add("format", "sdk");
            request.Parameters.Add("pretty", "true");

            if (publicRequest.FilterQuery != null)
                request.Parameters.Add("fq", StringUtils.FromString(publicRequest.FilterQuery));

            if (publicRequest.Query != null)
                request.Parameters.Add("q", StringUtils.FromString(publicRequest.Query));

            if (publicRequest.QueryOptions != null)
                request.Parameters.Add("q.options", StringUtils.FromString(publicRequest.QueryOptions));

            if (publicRequest.QueryParser != null)
                request.Parameters.Add("q.parser", StringUtils.FromString(publicRequest.QueryParser));

            if (publicRequest.Return != null)
                request.Parameters.Add("return", StringUtils.FromString(publicRequest.Return));

            if (publicRequest.Facet != null)
                request.Parameters.Add("facet", StringUtils.FromString(publicRequest.Facet));

            if (publicRequest.Sort != null)
                request.Parameters.Add("sort", StringUtils.FromString(publicRequest.Sort));


            request.Parameters.Add("size", StringUtils.FromLong(publicRequest.Size));

            request.ResourcePath = uriResourcePath;
            request.UseQueryString = false;
            request.SetContentFromParameters = true;

            return request;
        }

It does have some issues in that I've not added all the available properties to the request, but it does the job, and means that we can continue using the SDK without resorting to RestSharp.

I've copied most the code from your source, and traced the SDK code. I haven't had any issues with it yet.

[jimfl]

Added to https://github.com/aws/aws-sdk-net/blob/master/FEATURE_REQUESTS.md