changelog.md

ehttpc changes

0.7.0

• Switch to gun 2.1.0.
• Drop support for Erlang/OTP 25 and earlier.

0.6.0

• Changed log format to be more structured, host and port are included in the log data fields.
• Add max_inactive duration option (default is 10_000 milliseconds). This is to detect zombie connections especially when pipelining is set > 1. With {max_inactive, 10_000} added to the start_pool option, it will try to reconnect HTTP server up on detection of the last sent request had been expired for 10 seconds.

0.5.0

• Dropped hot-upgrade support.

• Added support for using HTTP proxy (HTTP 1.1 only). To use it, pass proxy in the pool opts.

  Ex:

  %% Point to the proxy host and port
  ProxyOpts = #{host => "127.0.0.1", port => 8888}.
  ehttpc_sup:start_pool(<<"pool">>, [{host, "target.host.com"}, {port, 80}, {proxy, ProxyOpts}]).
  
  %% To use username and password
  ProxyOpts = #{host => "127.0.0.1", port => 8888, username => "proxyuser", password => "secret"}.
  ehttpc_sup:start_pool(<<"pool">>, [{host, "target.host.com"}, {port, 80}, {proxy, ProxyOpts}]).

0.4.14

• Forcefully recreate gproc_pools during ehttpc_pool:init/1 to prevent reusing pools in an inconsistent state.

0.4.13

• Upgrade to gun 1.3.11. Fix host header

0.4.12

• Upgrade to gun 1.3.10 (OTP 26)
• Changed from gun:open to gun:start_link to start the gun process. This makes the gun process linked to ehttpc process (instead of gun_sup). Prior to this change, some early errors causing gun process to crash might not be able to be caught by ehttpc due to the slow monitoring. e.g. when some SSL option is invalid, the gun process will crash immediately and ehttpc can only get a noproc error reason from the delayed process monitor.

0.4.11

• Added support for PATCH method requests.

0.4.10

• Fixed ehttpc:request and ehttpc:request_async to handle Timeout = infinity option.

0.4.9

• Expanded the fix from 0.4.8 to also account for {error, {shutdown, normal}} return values in requests, and added a similar retry when the health check also fails with those two reasons.

0.4.8

• Fix an issue where a race condition would yield an {error, normal} return value. This can be caused when the gun process terminates when the remote server closes the connection for whatever reason. In this case, we simply retry without consuming "retry credits".

0.4.7

• Fix crash when using body headers and no body. When one sent a message with, for example, the content-type header and no body, the underlying gun library would expect a body to come with a gun:data/3 call and would not finish the request. The next request would then crash. See the following issue for more information: ninenines/gun#141

0.4.6

• Fix a bug introduced in 0.4.5: badarg crash from grpoc when pool_type is hash.

0.4.5

• Make possible to start pool with any() name (not limited to atom()).

0.4.4

• Fix the issue that ehttpc opens a new client when an existing client has been opened but not ready (no gun_up received).

0.4.3

• Update gproc dependency to 0.9.0.
• Add format_status/1 for OTP 25.

0.4.2

• Do not crash on retry and retry_timeout options. They might be persisted somehere, e.g. in an old config, or supervisor's child spec etc.

0.4.1

• Avoid sending stream cancellation messages to the 'gun' process if the stream is already finished ('fin' received)
• Force kill pool workers after 5 seconds waiting for graceful shutdown

0.4.0

• Add async APIs

0.3.0

• Changes on top of 0.2.1:

  • Prohibit the retry and retry_timeout options.

0.2.1

• Improvements and Bug Fixes

  • Add ehttpc:health_check/2.

0.2.0

• Major refactoring on top of 0.1.15
  • Added test cases.
  • Support hot upgrade from all 0.1.X versions.
  • No lower level retry (in 'gun' the http client lib).
  • Now enable_pipelining can be an integer to indicate the number of HTTP requests can be sent on wire before receiving responses (like the inflight-window). enable_pipelining=true has the same effect as enable_pipelining=100 and enable_pipelining=false has the same effect as enable_pipelining=1
  • Now all requests are async, so the ehttpc:request calls can be collected from mailbox into process state, this makes the handling of gun responses more effecient.

0.1.14

• fixed appup. old versions missed ehttpc_pool
• added check_vsn.escript to sure version consistent and run in ci