I understand why the following snippet doesn't work in Firefox and in Chrome: we're making an AJAX request to another domain.
var xhr = new XMLHttpRequest();
xhr.open("GET", "http://www.perdu.com", true);
xhr.addEventListener("load", function() { console.debug(xhr.responseText); }, false);
xhr.send(null);
But then why does Safari output this? This is the actual content of the page. I have Safari 7.1.
<html><head><title>Vous Etes Perdu ?</title></head><body><h1>Perdu sur l'Internet ?</h1><h2>Pas de panique, on va vous aider</h2><strong><pre> * <----- vous êtes ici</pre></strong></body></html>
It turns out that Safari behaves differently when loading from a server or from a file system.
The original answer below tests the CORS functionality with a file:///
scheme. Safari lets users bypass CORS on that scheme.
As Jonathan Crowe pointed out on localhost
, so with the http://
scheme, Safari blocks the response, same as Firefox and Chrome.
So there is no bug on this one. As for the behaviour on the file system, I guess we can call it a feature, or a convenience (thinking about quick local tests)?
Note: This update relies on an additional, simple test to serve the HTML snippet below from an HTTP server. Nothing fancy, and Safari just behaves as the others.
I could reproduce the problem. It might be a bug in Safari 7.1, and here is what I found toward that temporary conclusion.
Origin
header. Others set it to null
.Also, this version of Safari allows setting the Origin
header on the XMLHttpRequest
object (Chrome does not):
xhr.setRequestHeader("Origin", null);
Setting the header to null
to get closer to the other browsers does not change the result: Safari 7.1 still allows the response to get through to the requester.
I could not make sure this is a bug in Safari 7.1, but it seems to be its behaviour right now.
Some details below.
<html>
<script>
var xhr = new XMLHttpRequest();
xhr.open("GET", "http://www.perdu.com", true);
xhr.addEventListener("load", function() { console.debug(xhr.responseText); }, false);
xhr.send(null);
</script>
</html>
(All on Mac OS X 10.9.5)
GET / HTTP/1.1
Host: www.perdu.com
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:33.0) Gecko/20100101 Firefox/33.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Origin: null
Connection: keep-alive
GET / HTTP/1.1
Host: www.perdu.com
Connection: keep-alive
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/38.0.2125.122 Safari/537.36
Origin: null
Accept: */*
DNT: 1
Accept-Encoding: gzip,deflate,sdch
Accept-Language: en-US,en;q=0.8,fr;q=0.6,ja;q=0.4,pt;q=0.2
GET / HTTP/1.1
Host: www.perdu.com
Accept-Encoding: gzip, deflate
Accept: */*
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_5) AppleWebKit/600.1.17 (KHTML, like Gecko) Version/7.1 Safari/537.85.10
Accept-Language: en-us
DNT: 1
Connection: keep-alive
I am currently testing edge cases with CORS over a range of browsers for a web API. If a bug gets confirmed, it shouldn't be too much of a problem---provided the API security is serious enough (as CORS does not secure the server) !
I have asked Apple if they can confirm on their feedback site.