[Openmcl-devel] Hunchentoot and CCL on Windows XP

Gary Byers gb at clozure.com
Thu Mar 19 17:46:13 UTC 2009


If I wasn't clear: the bug's in CCL on Windows (we should
either implement the timeouts correctly or ignore them),
not in Edi's code.

On Thu, 19 Mar 2009, Sean Ross wrote:

>
> On 19 Mar 2009, at 16:34, Gary Byers wrote:
>
>> Thanks for the backtrace.
>> 
>> It's a bug, triggered by the facts that:
>> 
>> 1) the socket is created with a non-null :INPUT-TIMEOUT and/or
>>  :OUTPUT-TIMEOUT argument
>> 
>> 2) that isn't handled correctly (in several ways) on Windows.
>> 
>> If you can find the call(s) to MAKE-SOCKET that set those arguments
>> and conditionalize them out:
>> 
>> (make-socket ... #+(and ccl (not windows-target)) :input-timeout ...)
>> 
>> until that works, you should at least get past that.
>> 
>
>
> Thanks Gary,
> Thats sorted the problem, I'll forward on a patch to Edi.
>
>
>
> ----
>
> For reference purposes (and google juice) you can get this all working on 
> Windows by doing
>
> (hunchentoot:start (make-instance 'hunchentoot:acceptor :read-timeout nil 
> :write-timeout nil :port 8080))
>
>



More information about the Openmcl-devel mailing list