[hunchentoot-devel] Hard to catch errors below process-request

[Hans Hübner]

Hi Andrei,

thank you for sending the reproduction script.  Fixing the problem in
a satisfactory fashion requires some refactoring that'll take me some
time and require review.  I'll keep you posted.

-Hans

On Fri, Jan 21, 2011 at 5:15 PM, AVS <lispercat at gmail.com> wrote:
> Here is the standalone file to reproduce the unicode failure problem
> on my system (for the code below, put your path to quicklisp and save
> the code in a test.lisp file ):
> Load the file (for sbcl it's "sbcl --load test.lisp"), run lynx and
> type in "lynx 127.0.0.1:1234/test.html". It should give you an empty
> page.
>
> ;;test.lisp
> (load "/your-path/quicklisp/setup.lisp")
> (ql:quickload "hunchentoot")
>
> (defpackage :test
>  (:use :cl :hunchentoot))
>
> (in-package :test)
>
> (defparameter *server* (hunchentoot:start (make-instance
> 'hunchentoot:acceptor :port 1234)))
> (hunchentoot:define-easy-handler (test :uri "/test.html") (name)
>  (coerce '(#\RIGHT_DOUBLE_QUOTATION_MARK #\t #\e #\s #\t
> #\RIGHT_DOUBLE_QUOTATION_MARK) 'string))
> ;;file end
>
> Thank you,
> Andrei
>
>
> On Fri, Jan 21, 2011 at 5:05 AM, Hans Hübner <hans.huebner at gmail.com> wrote:
>> Harak, Andrei,
>>
>> I had a look at the relevant positions in the Hunchentoot source and
>> would be interested in solving the problem so that errors are
>> reported, rather than swallowed, without requiring the user to set up
>> special handlers.  My time is limited, though, so if you could help me
>> by providing an executable test case that exposes the problem, it´d be
>> greatly appreciated.
>>
>> Thanks,
>> Hans
>>
>> On Fri, Jan 21, 2011 at 10:48 AM, Phil Marneweck <haragx at gmail.com> wrote:
>>> On Fri, 2011-01-21 at 06:40 +0100, Hans Hübner wrote:
>>>
>>> On Fri, Jan 21, 2011 at 4:29 AM, AVS <lispercat at gmail.com> wrote:
>>>> Thank you for the suggestion, Hans.
>>>> In the mean time could you try one thing? Put some bad code inside
>>>> start-output that will cause an exception (for instance (list a b)
>>>> without a, b defined, or (/ 1 0)). Also (setf *catch-errors-p* nil).
>>>> If things are as they should be it'll trigger the debugger. In my case
>>>> it doesn't which is the cause of the whole issue.
>>>> If it triggers the debugger break for you, it means there is something
>>>> related to sbcl I am using.
>>>
>>> The debugger is not triggered, so the bug is real.  Having a way to
>>> reproduce it by user code would be helpful nevertheless.
>>>
>>> -Hans
>>>
>>>
>>> Like following?
>>>
>>> (defpackage :test
>>>   (:nicknames :test )
>>>   (:use :cl :hunchentoot :cl-who) )
>>>
>>>
>>> (in-package :test)
>>>
>>> ;;(hunchentoot:start (make-instance 'hunchentoot:acceptor :port 1111 ))
>>>
>>>
>>> (hunchentoot::define-easy-handler  (eish :uri "/eish.html"
>>> :default-request-type :both)
>>> ()
>>>    (cl-who::with-html-output-to-string (*standard-output* nil :indent t)
>>>      (cl-who::str "Hello World  XXX")))
>>>
>>>
>>> (defmacro with-debugging (&body body)
>>>   ;; Using this as debugging tool because hunchentoot
>>>   ;; swallows all errors here.
>>>   `(handler-bind ((error #'invoke-debugger))
>>>      , at body))
>>>
>>> (defclass xxx ()
>>>   ((ccc)) )
>>>
>>>
>>> (defmethod handle-request :before ((*acceptor* hunchentoot::acceptor)
>>> (*request* hunchentoot::request))
>>>
>>> ; (with-debugging
>>>    (slot-value (make-instance 'xxx) 'xxx)
>>> ;)
>>> )
>>>
>>> Harag
>>> _______________________________________________
>>> tbnl-devel site list
>>> tbnl-devel at common-lisp.net
>>> http://common-lisp.net/mailman/listinfo/tbnl-devel
>>>
>>
>> _______________________________________________
>> tbnl-devel site list
>> tbnl-devel at common-lisp.net
>> http://common-lisp.net/mailman/listinfo/tbnl-devel
>>
>
> _______________________________________________
> tbnl-devel site list
> tbnl-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/tbnl-devel
>