mirror of
https://github.com/puma/puma.git
synced 2022-11-09 13:48:40 -05:00
2deaf77a38
* Swallow connection errors when sending early hints I've been thinking on how Puma should behave when the early hints can't be sent because the connection has been lost. I first thought that it should behave the same way it does when the connection is lost before main response is sent, I've seen that it swallows those errors (https://git.io/fjVtg) so perhaps we should do the same here. There's one consideration though. Because the early hints are sent at the very beginning of the app execution, we may want to abort this request altogether and save resources by throwing an exception and causing Rails to interrupt execution. Finally I decided to overlook that case just because I believe it's least surprising from the Rack app point of view. Also I haven't seen any other example of the web server interrupting the apps execution due to an error on the lower layer. Fixes #1640. * Adds a test * Updates test as per @nateberkopec's suggestion |
||
|---|---|---|
| .. | ||
| app | ||
| plugin | ||
| rack | ||
| accept_nonblock.rb | ||
| binder.rb | ||
| cli.rb | ||
| client.rb | ||
| cluster.rb | ||
| commonlogger.rb | ||
| configuration.rb | ||
| const.rb | ||
| control_cli.rb | ||
| convenient.rb | ||
| daemon_ext.rb | ||
| delegation.rb | ||
| detect.rb | ||
| dsl.rb | ||
| events.rb | ||
| io_buffer.rb | ||
| jruby_restart.rb | ||
| launcher.rb | ||
| minissl.rb | ||
| null_io.rb | ||
| plugin.rb | ||
| rack_default.rb | ||
| reactor.rb | ||
| runner.rb | ||
| server.rb | ||
| single.rb | ||
| state_file.rb | ||
| tcp_logger.rb | ||
| thread_pool.rb | ||
| util.rb | ||