From: matthew@... Date: 2014-06-10T19:58:48+00:00 Subject: [ruby-core:63079] [ruby-trunk - Bug #9927] webrick does not unset content-length when responding to HEAD requests Issue #9927 has been updated by Matthew Kerwin. Adrien Thebo wrote: > I didn't make this very clear, but the `curl` invocation hangs for about 6 seconds until the webrick server is killed, and then prints '`transfer closed with 23 bytes remaining to read`' which indicates it was hanging. This is a problem with the way you're using curl. Webrick is doing the right thing according to the HTTP spec[1], which states that "the payload header fields [including content-length] MAY be omitted." There is no obligation on the server to do so. You have to tell your client (curl) not to wait for a message body. -X HEAD just changes bytes omitted in the request, but doesn't change the client behaviour; you have to use -I to signal that this is a metadata-only fetch [2] [1] http://tools.ietf.org/html/rfc7231#section-4.3.2 [2] http://linux.die.net/man/1/curl ---------------------------------------- Bug #9927: webrick does not unset content-length when responding to HEAD requests https://bugs.ruby-lang.org/issues/9927#change-47154 * Author: Adrien Thebo * Status: Rejected * Priority: Normal * Assignee: * Category: * Target version: * ruby -v: ruby 1.9.3p484 (2013-11-22 revision 43786) [x86_64-linux] * Backport: 2.0.0: UNKNOWN, 2.1: UNKNOWN ---------------------------------------- When Webrick responds to HEAD requests it omits the body (per RFC2616 -- 4.4 Message Length). However when setting up the response headers the content-length field is set to the length of the body, which means that the resulting response will have a content length that doesn't match the actual response. This means that some HTTP clients may hang when reading the response. This is reproducible with the following: ```ruby require 'webrick' server = WEBrick::HTTPServer.new :Port => 8080, :BindAddress => '127.0.0.1' server.mount_proc("/") do |req, res| res.body = "This will be ignored!\r\n" end trap('INT') do server.shutdown end server.start ``` Running this with curl results in the following: ~~~ ���> ruby webrick-head.rb [2014-06-10 12:07:28] INFO WEBrick 1.3.1 [2014-06-10 12:07:28] INFO ruby 1.9.3 (2013-11-22) [x86_64-linux] [2014-06-10 12:07:28] INFO WEBrick::HTTPServer#start: pid=24798 port=8080 localhost - - [10/Jun/2014:12:07:30 PDT] "HEAD / HTTP/1.1" 200 0 - -> / ^C[2014-06-10 12:07:36] INFO going to shutdown ... [2014-06-10 12:07:36] INFO WEBrick::HTTPServer#start done. ~~~ ~~~ ���> curl -v -X HEAD http://localhost:8080 * Rebuilt URL to: http://localhost:8080/ * Hostname was NOT found in DNS cache * Trying 127.0.0.1... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0* Connected to localhost (127.0.0.1) port 8080 (#0) > HEAD / HTTP/1.1 > User-Agent: curl/7.37.0 > Host: localhost:8080 > Accept: */* > < HTTP/1.1 200 OK * Server WEBrick/1.3.1 (Ruby/1.9.3/2013-11-22) is not blacklisted < Server: WEBrick/1.3.1 (Ruby/1.9.3/2013-11-22) < Date: Tue, 10 Jun 2014 19:07:30 GMT < Content-Length: 23 < Connection: Keep-Alive < 0 23 0 0 0 0 0 0 --:--:-- 0:00:01 --:--:-- 0 0 23 0 0 0 0 0 0 --:--:-- 0:00:02 --:--:-- 0 0 23 0 0 0 0 0 0 --:--:-- 0:00:03 --:--:-- 0 0 23 0 0 0 0 0 0 --:--:-- 0:00:04 --:--:-- 0 0 23 0 0 0 0 0 0 --:--:-- 0:00:05 --:--:-- 0 [Ctrl-C sent to server] 0 23 0 0 0 0 0 0 --:--:-- 0:00:06 --:--:-- 0{ [data not shown] * transfer closed with 23 bytes remaining to read 0 23 0 0 0 0 0 0 --:--:-- 0:00:06 --:--:-- 0 * Closing connection 0 curl: (18) transfer closed with 23 bytes remaining to read ~~~ This is reasonably straightforward to fix; when the headers are being created and the code is checking to see if the body should be ignored for HTTP 204 and 304, we can check to see if we're responding to a HEAD request and behave accordingly. I've attached patches to this effect ---Files-------------------------------- 0001-lib-webrick-httpresponse.rb-unset-content-length-hea.patch (1.54 KB) -- https://bugs.ruby-lang.org/