Wikimedia Bugzilla is closed!

I'm struggling to reproduce this.

<?php
header( "ERROR: I'm a teapot!", true, 418 );
echo 'teapot!';

See below. Output from execute is hacked in on the read....


> $req = MWHttpRequest::factory( "http://192.168.0.195/teapot.php", $options );

> $status = $req->execute();
teapot!
> var_dump( $status );
object(Status)#14 (7) {
  ["ok"]=>
  bool(false)
  ["value"]=>
  int(100)
  ["successCount"]=>
  int(0)
  ["failCount"]=>
  int(0)
  ["success"]=>
  array(0) {
  }
  ["errors"]=>
  array(1) {
    [0]=>
    array(3) {
      ["type"]=>
      string(5) "error"
      ["message"]=>
      string(15) "http-bad-status"
      ["params"]=>
      array(2) {
        [0]=>
        string(3) "418"
        [1]=>
        string(6) "unused"
      }
    }
  }
  ["cleanCallback"]=>
  bool(false)
}

> var_dump( $req->getContent() );
string(7) "teapot!"

I also can not reproduce (tried with 404 page).

The Http::get wrapper won't return content for non-200 pages it seems, but MWHttpRequest::factory seems to work fine.

Created attachment 8299 [details]
patch to still get response body for non-200 response codes

After discussion on irc, turns out the issue is in PhpHttpRequest, not CurlHttpRequest.

Here's a patch that should in theory fix it, and does fix it for Nikerabbit, but doesn't work for me :(

I'm not entirely sure whats going on, but I get http-read-errors (aka $fh is null) for any request that isn't a 200.

When I tried on TW, I got returned an instance of CurlHttpRequest

Weird.

Well, the Curl works now on twn, I don't know why it didn't earlier (I maybe the server needs multiple restarts for it to take action). I switched manually to php version to test the patch.

That's possible, I know the webservers need bouncing after PHP modules etc are added...

If it works, it might aswell be applied minus the var_dump() and closed...

Applied minus var_dump() in r84168