Hey, I just attached a possible patch for JRUBY-3812, which reports
that we're doing a second reverse DNS lookup whenever you
Socket.gethostbyname. The problem stems from the way Java's
InetAddress logic works. We do the forward lookup, and JDK constructs
an InetAddress object without hostname information. Then we ask it to
getCanonicalHostName, and since it doesn't have a name it has to do a
second reverse lookup.

The fix I made was to just use the original requested hostname as the
one we return. It doesn't appear to break any tests, but I still
wonder if it's correct. Do we need the "canonical" host name? Does it
matter if we just reuse the requested hostname?

Have a look...let me know what you think:
http://jira.codehaus.org/browse/JRUBY-3812

- Charlie

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email


Reply via email to