http2

Checklist

• Dependencies installed
• No typos
• Searched existing issues and docs

Issue Description

When using the RateLimiter Middleware with a rate between 0 and 1 all events will be rejected instead of applying the specified rate. E.g.: e.Use(middleware.RateLimiter(middleware.NewRateLimiterMemoryStore(0.5)))

I am not saying that it is a common use case to have

When sending a request body that decodes a Content type. If the request is sent as multipart with the right header set, but the body is empty, a 500 error is returned. This can cause problems with apps that have automatic reporting for internal server errors as it should be reported as a bad request

The MinSendBackupAfterMs is now set to 1 ms. Unfortunately this causes backup requests to be send when there is low load. It would be useful if we could set the minimum a little bit higher for certain endpoints.

Latest Sofa Hessian version is 3.3.13, we can update it in Sofa rpc

If a user performs a blocking task in decorator chains or services, they have to 1) submit a task to a blocking task executor,
2) pass the result to a CompletableFuture and 3) wrap the CompletableFuture with HttpResponse.from().

sb.serviceUnder("/web", service.decorate((delegate, ctx, req) -> {
    CompletableFuture<HttpResponse> future = new CompletableFuture<>();
    ctx.bl

Currently we have a mix of:

• Out of memory error.
• Out of memory error
• out of memory error
• out of memory
• out of memory.
• out of memory [some context]

All these should be adjusted to use the same wording for consistency and it should also make the executable a little bit smaller 😄