🐛 Bug Report

We shouldn't send trailers unless the client sends the "TE" trailer.

See grpc-ecosystem/grpc-gateway#1697 for more context

我需要

可我在hyperf中不知道该如何使用write

In our codebase, we push RequestContext and immediately run some code with try-with-resources.
For example:

try (SafeCloseable ignored = ctx.push()) {
    logger.trace(decorate(msg));
}

If RequestContext provides run(Runnable) or call(Callable) we can reduce boilerplate code and simplify it.

ctx.run(() -> logger.trace(decorate(msg));

This is inspired by g

We have a fancy logo from the new tokio website, we should go through the code base and update it in every spot we use the old one.

https://github.com/tokio-rs/website/blob/master/public/img/icons/tonic.svg

virtual host options

Issue Description

是否有计划用actions 完全取代 TravisCI
感觉actions 速度上比 TravisCI更好点 ，另外也是github原生的。

Type: feature request

Describe what feature

Feature Request

Implement something similar to UNIX du command. Name might be something else, e.g. talosctl usage (with alias du).

Description

Based on existing List() API, aggregate size by directories, count size of all elements, support --depth flag.

At present we can run the load generator for a set amount of time or a certain number of requests.
I would like to a mode --reconnetRequests=1000 where fortio client will reconnect after 1000 requests.

Use case: We are testing SDS cert rotation using an https fortio client (std client).
For cert rotation to be exercised fortio should reconnect periodically to fetch the new cert.
In