On 06/20/2013 01:57 PM, John Zavgren wrote:
Chris: Your idea of moving the empty string check so that it's header specific is prudent... that's less likely to have unforeseen consequences. I'll make the change ASAP.
and maybe a test? -Chris.
John On 06/20/2013 05:56 AM, Chris Hegarty wrote:Thanks John, I just did a quick test with the testcase attached to the bug report ( below), and the server is indeed replying with a Set-Cookie header with no value ( treated as empty string ) <sigh!> Since all header retrieval passes through filterHeaderField, in one way or another, I'm a little concerned about changing this. Also, as the only issue we know of is with Set-Cookie(2), maybe you could add the empty string check to these headers only? ( that is to say, move the 'value.length() == 0' check into the ' if (SET_COOKIE.equalsIgnoreCase(name)..... ' ---- public class EmptyCookieHeaderStringTest { public static void main(String[] args) throws IOException { CookieHandler.setDefault(new TestCookieHandler()); URL url = new URL("http://www.walmart.com/msharbor/";); HttpURLConnection c = (HttpURLConnection) url.openConnection(); c.getHeaderFields(); } } class TestCookieHandler extends CookieHandler { @Override public Map<String, List<String>> get(URI uri, Map<String, List<String>> requestHeaders) { return new HashMap<String, List<String>>(); } @Override public void put(URI uri, Map<String, List<String>> responseHeaders) { } } ----- -Chris. On 06/19/2013 04:27 PM, John Zavgren wrote:Greetings: Please review the following changes for a bug that's caused by empty cookie header strings. The proposed change detects this condition and returns "early". http://cr.openjdk.java.net/~jzavgren/8015799/webrev.01/ <http://cr.openjdk.java.net/%7Ejzavgren/8015799/webrev.01/> -- John Zavgren john.zavg...@oracle.com 603-821-0904 US-Burlington-MA
