Loading ...
Sorry, an error occurred while loading the content.

Re: on_fault

Expand Messages
  • Ray Rizzuto
    Paul, There s no perfect solution to error handling since it depends so much on the application. In my case, I am choosing to treat transport errors as
    Message 1 of 9 , Feb 14, 2001
    • 0 Attachment
      Paul,

      There's no perfect solution to error handling since it depends so
      much on the application. In my case, I am choosing to treat
      transport errors as non-fatal in the belief that they can be
      retried. Application errors (faults?) hopefully don't happen, but
      when they do they probably require manual intervention.

      I think I can almost get the behavior I want with an appropriate
      on_fault handler. I have a few questions/suggestions:

      1) could you give more details on the parameters passed to the fault
      handler? And also what the fault handler can return and how that is
      used? It seems that what it returns is returned to the failed call,
      but I'm not positive.

      2) in one of the examples you provide, you are creating a new
      SOAP::SOM to return from the fault handler:

      use SOAP::Lite
      on_fault => sub { my($soap, $res) = @_;
      eval { die ref $res ? $res->faultdetail : $soap->transport-
      >status };
      return ref $res ? $res : new SOAP::SOM;
      };

      If I could set the fault values in this new SOM, I would be able to
      address errors uniformly after the original call. Or am I
      misunderstanding this code?

      3) in the above code, on a transport error @_ has 2 entries, but the
      second one seems to just have an empty string. Is that the defined
      behavior for a transport error?

      Ray

      --- In soaplite@y..., Paul Kulchenko <paulclinger@y...> wrote:
      > Hi, Ray!
      >
      > That is perfectly right and I understand it, but for several reasons
      > default value was chosen as die on transport and return fault on
      SOAP
      > errors. You may easily alter this behavior (you may even specify it
      > once for ALL objects in your code with GLOBAL SETTINGS), but there
      is
      > no one-size-fit-all solution.
      >
      > I thought about doing styles, so you can choose for example, die
      > style and I'll install handler for you, or warning style or error
      > code style or whatever, but i think it's too complicated conception.
      >
      > Best wishes, Paul.
      >
      > --- Ray Rizzuto <ray.rizzuto@u...> wrote:
      > > Paul,
      > >
      > > I'd like to be able to handle the transport errors in the same way
      > > as
      > > faulted calls, via the fault methods of the SOAP::SOM returned
      from
      > >
      > > the call. That's why I was trying to create a SOAP::SOM and set
      > > the
      > > string.
      > >
      > > Part of the reason for this is that transport errors may be
      > > transient - I'd like to retry the call later in the application
      I'm
      > >
      > > writing. My app is a gui interface to a problem tracking system.
      > > If
      > > the user is creating or editing a problem report, and the network
      > > goes down, I want them to be able to reattempt the update in a
      few
      > > minutes.
      > >
      > > Ray
      > >
      > >
      > >
      > > --- In soaplite@y..., Paul Kulchenko <paulclinger@y...> wrote:
      > > > Hi, Ray!
      > > >
      > > > --- Ray Rizzuto <ray.rizzuto@u...> wrote:
      > > > > I'd like to be able to handle all errors from a soap call,
      > > > > including
      > > > > transport errors, after the call completes. I've set the
      > > following
      > > > > fault handler, which almost works except that the faultstring
      > > isn't
      > > > > available after the call to the method:
      > > > >
      > > > > sub fault
      > > > > {
      > > > > my($soap, $res) = @_;
      > > > > if (defined $res && ref $res)
      > > > > {
      > > > > print $res->faultstring;
      > > > > return $res;
      > > > > }
      > > > > else
      > > > > {
      > > > > print $soap->transport->status;
      > > > > $res = new SOAP::SOM;
      > > > > $res->faultstring("transport error");
      > > >
      > > > Here is the reason. faultstring() and other methods are
      readonly.
      > >
      > > You
      > > > may get access to message structure, but you cannot
      update/modify
      > >
      > > it.
      > > > Next version probably will generate warning/error if you try to
      > > > provide parameter(s) for these methods.
      > > >
      > > > > I'm very new to this, so I may be making an obvious mistake.
      > > >
      > > > You may check Error Handling section on
      > > http://guide.soaplite.com/.
      > > I
      > > > prefer 2.k example and in any case you always have access to
      > > result
      > > > of the call through call() method:
      > > >
      > > > print $soap->mymethod(123)->result;
      > > > print $soap->call->faultstring;
      > > >
      > > > Even if call made with autodispatch:
      > > >
      > > > use SOAP::Lite +autodispatch => ...;
      > > >
      > > > print mymethod(123);
      > > > print SOAP::Lite->self->call->faultstring;
      > > >
      > > > Best wishes, Paul.
      > > >
      > > > __________________________________________________
      > > > Do You Yahoo!?
      > > > Get personalized email addresses from Yahoo! Mail - only $35
      > > > a year! http://personal.mail.yahoo.com/
      > >
      > >
      > > ------------------------ Yahoo! Groups Sponsor
      > >
      > > To unsubscribe from this group, send an email to:
      > > soaplite-unsubscribe@y...
      > >
      > >
      > >
      >
      >
      > __________________________________________________
      > Do You Yahoo!?
      > Get personalized email addresses from Yahoo! Mail - only $35
      > a year! http://personal.mail.yahoo.com/
    • Paul Kulchenko
      Hi, Ray! ... Unfortunately not only. It may depend on Server implementation also, though I tried hard to hide these details from you. Several situations are
      Message 2 of 9 , Feb 14, 2001
      • 0 Attachment
        Hi, Ray!

        --- Ray Rizzuto <ray.rizzuto@...> wrote:
        > There's no perfect solution to error handling since it depends so
        > much on the application. In my case, I am choosing to treat
        Unfortunately not only. It may depend on Server implementation also,
        though I tried hard to hide these details from you.
        Several situations are possible:
        1. 200 OK + SOAP message (normal result)
        2a. 500 Server Error + error message (not SOAP)
        2b. 400 Bad Method + error message (not SOAP)
        2c. 510 Not Extended + error message (not SOAP)
        3a. 500 Server Error + SOAP message (Fault result)
        3b. 400 Bad Method + SOAP message (Fault result)
        4. 200 OK + SOAP message (Fault result)

        2c will be handled by SOAP::Lite automatically (you won't see it, as
        well as redirect from server).
        3b and 4 are not allowed according to current spec., though there was
        long debates about using 1 and 4 instead of 1 and 3a.

        > 1) could you give more details on the parameters passed to the
        > fault handler? And also what the fault handler can return and how
        > that
        > is used? It seems that what it returns is returned to the failed
        > call, but I'm not positive.

        First, on_fault will be called unless transport returned success.
        on_fault accepts two parameters: SOAP::Lite object and deserialized
        result (exactly as you would get it after $soap->mymethod() call).
        You may return this deserialized result or create your own and it'll
        become the result of call.
        In case you returned false (on_fault(sub{})) result of the call will
        be deserialized message as if on_fault wasn't call at all (no error
        handling).

        > 2) in one of the examples you provide, you are creating a new
        > SOAP::SOM to return from the fault handler:
        >
        > use SOAP::Lite
        > on_fault => sub { my($soap, $res) = @_;
        > eval { die ref $res ? $res->faultdetail : $soap->transport-
        > >status };
        > return ref $res ? $res : new SOAP::SOM;
        > };
        >
        > If I could set the fault values in this new SOM, I would be able to
        > address errors uniformly after the original call. Or am I
        > misunderstanding this code?
        No, you're not. You cannot assign it, but you may create new SOM
        object as result of deserialization:

        use SOAP::Lite
        on_fault => sub { my($soap, $res) = @_;
        eval { die ref $res ? $res->faultdetail : $soap->transport->status
        };
        return ref $res ? $res :
        $soap->deserializer->deserialize(
        $soap->serializer->fault('Server.Transport',
        $soap->transport->status));
        };

        print SOAP::Lite->proxy('http://localhost/')->mymethod->faultstring;

        > 3) in the above code, on a transport error @_ has 2 entries, but
        > the
        > second one seems to just have an empty string. Is that the defined
        > behavior for a transport error?
        There should be deserialized message (if you got Fault response) OR
        content of the message (if message couldn't be deserialized, for
        example HTML code or some string with error message), so if you check
        and there is not refrence, you can be sure that it's pure transport
        error (and not SOAP Fault).

        Hope it helps.

        Best wishes, Paul.


        __________________________________________________
        Do You Yahoo!?
        Get personalized email addresses from Yahoo! Mail - only $35
        a year! http://personal.mail.yahoo.com/
      • Ray Rizzuto
        Paul, In your list of errors, I don t see the case I m experiencing, which is a timeout: 500 Can t connect to pc-rizzuto-p:9999 (Timeout) In that case, what
        Message 3 of 9 , Feb 14, 2001
        • 0 Attachment
          Paul,
           
          In your list of errors, I don't see the case I'm experiencing, which is a timeout:
           
          500 Can't connect to pc-rizzuto-p:9999 (Timeout)
           
          In that case, what should I get as the second parameter to the fault handler?  In the debugger, @_ has length 2, and print $_[1], but print ref $_[1] returns an empty string.  Is that because the content of the response is empty?
           
          Ray
           
          P.S.  Thanks for the information on using $soap->deserializer->deserialize to create a SOM with the appropriate fault information in it.  I would never have been able to figure that out by myself!
          -----Original Message-----
          From: Paul Kulchenko [mailto:paulclinger@...]
          Sent: Wednesday, February 14, 2001 12:46 PM
          To: soaplite@yahoogroups.com
          Subject: Re: [soaplite] Re: on_fault

          Hi, Ray!

          --- Ray Rizzuto <ray.rizzuto@...> wrote:
          > There's no perfect solution to error handling since it depends so
          > much on the application.  In my case, I am choosing to treat
          Unfortunately not only. It may depend on Server implementation also,
          though I tried hard to hide these details from you.
          Several situations are possible:
            1. 200 OK + SOAP message (normal result)
            2a. 500 Server Error + error message (not SOAP)
            2b. 400 Bad Method + error message (not SOAP)
            2c. 510 Not Extended + error message (not SOAP)
            3a. 500 Server Error + SOAP message (Fault result)
            3b. 400 Bad Method + SOAP message (Fault result)
            4. 200 OK + SOAP message (Fault result)

          2c will be handled by SOAP::Lite automatically (you won't see it, as
          well as redirect from server).
          3b and 4 are not allowed according to current spec., though there was
          long debates about using 1 and 4 instead of 1 and 3a.

          > 1) could you give more details on the parameters passed to the
          > fault handler? And also what the fault handler can return and how
          > that
          > is used?  It seems that what it returns is returned to the failed
          > call, but I'm not positive.

          First, on_fault will be called unless transport returned success. 
          on_fault accepts two parameters: SOAP::Lite object and deserialized
          result (exactly as you would get it after $soap->mymethod() call).
          You may return this deserialized result or create your own and it'll
          become the result of call.
          In case you returned false (on_fault(sub{})) result of the call will
          be deserialized message as if on_fault wasn't call at all (no error
          handling).

          > 2) in one of the examples you provide, you are creating a new
          > SOAP::SOM to return from the fault handler:
          >
          >   use SOAP::Lite
          >     on_fault => sub { my($soap, $res) = @_;
          >       eval { die ref $res ? $res->faultdetail : $soap->transport-
          > >status };
          >       return ref $res ? $res : new SOAP::SOM;
          >     };
          >
          > If I could set the fault values in this new SOM, I would be able to
          > address errors uniformly after the original call.  Or am I
          > misunderstanding this code?
          No, you're not. You cannot assign it, but you may create new SOM
          object as result of deserialization:

          use SOAP::Lite
            on_fault => sub { my($soap, $res) = @_;
             eval { die ref $res ? $res->faultdetail : $soap->transport->status
          };
               return ref $res ? $res :
                 $soap->deserializer->deserialize(
                   $soap->serializer->fault('Server.Transport',
          $soap->transport->status));
             };

          print SOAP::Lite->proxy('http://localhost/')->mymethod->faultstring;

          > 3) in the above code, on a transport error @_ has 2 entries, but
          > the
          > second one seems to just have an empty string.  Is that the defined
          > behavior for a transport error?
          There should be deserialized message (if you got Fault response) OR
          content of the message (if message couldn't be deserialized, for
          example HTML code or some string with error message), so if you check
          and there is not refrence, you can be sure that it's pure transport
          error (and not SOAP Fault).

          Hope it helps.

          Best wishes, Paul.


          __________________________________________________
          Do You Yahoo!?
          Get personalized email addresses from Yahoo! Mail - only $35
          a year!  http://personal.mail.yahoo.com/


          To unsubscribe from this group, send an email to:
          soaplite-unsubscribe@yahoogroups.com


        • Paul Kulchenko
          Hi, Ray! ... Agree. Fixed list: OK Transport + OK SOAP 1. 200 OK + SOAP message (normal result) Error Transport + no SOAP 2a. 500 Server Error + error message
          Message 4 of 9 , Feb 14, 2001
          • 0 Attachment
            Hi, Ray!

            --- Ray Rizzuto <ray.rizzuto@...> wrote:
            > In your list of errors, I don't see the case I'm experiencing,
            > which is a timeout:
            >
            > 500 Can't connect to pc-rizzuto-p:9999 (Timeout)
            Agree. Fixed list:

            OK Transport + OK SOAP
            1. 200 OK + SOAP message (normal result)

            Error Transport + no SOAP
            2a. 500 Server Error + error message (not SOAP)
            2b. 400 Bad Method + error message (not SOAP)
            2c. 510 Not Extended + error message (not SOAP)
            2d. All other errors

            Error Transport + Fault SOAP
            3a. 500 Server Error + SOAP message (Fault result)
            3b. 400 Bad Method + SOAP message (Fault result)

            OK transport + Fault SOAP
            4. 200 OK + SOAP message (Fault result)

            > In that case, what should I get as the second parameter to the
            > fault
            > handler? In the debugger, @_ has length 2, and print $_[1], but
            > print ref
            > $_[1] returns an empty string. Is that because the content of the
            > response is empty?
            Exactly. ref $_[1] will return empty string even if $_[1] is not
            empty, but doesn't represent any referense and contains plain string,
            like 'abc'. That means that content of message either empty or not
            parsed.

            Best wishes, Paul.


            __________________________________________________
            Do You Yahoo!?
            Get personalized email addresses from Yahoo! Mail - only $35
            a year! http://personal.mail.yahoo.com/
          • Ray Rizzuto
            Paul, I had a minor typo. In a timeout situation, in the fault handler: print $_[1] prints an empty string print defined $_[1] prints 1 print ref $_[1]
            Message 5 of 9 , Feb 14, 2001
            • 0 Attachment
              Paul,
               
              I had a minor typo.  In a timeout situation, in the fault handler:
               
              print $_[1] prints an empty string
              print defined $_[1]  prints 1
              print ref $_[1]  prints an empty string
               
              Since this is a transport error $_[1] is string, which is why ref $_[1] is empty.  Since the content of the returned message is empty in the case of a timeout, that explains why $_[1] is also an empty string.
               
              Ray
              -----Original Message-----
              From: Paul Kulchenko [mailto:paulclinger@...]
              Sent: Wednesday, February 14, 2001 4:16 PM
              To: soaplite@yahoogroups.com
              Subject: RE: [soaplite] Re: on_fault

              Hi, Ray!

              --- Ray Rizzuto <ray.rizzuto@...> wrote:
              > In your list of errors, I don't see the case I'm experiencing,
              > which is a timeout:
              >
              > 500 Can't connect to pc-rizzuto-p:9999 (Timeout)
              Agree. Fixed list:

              OK Transport + OK SOAP
              1. 200 OK + SOAP message (normal result)

              Error Transport + no SOAP
              2a. 500 Server Error + error message (not SOAP)
              2b. 400 Bad Method + error message (not SOAP)
              2c. 510 Not Extended + error message (not SOAP)
              2d. All other errors

              Error Transport + Fault SOAP
              3a. 500 Server Error + SOAP message (Fault result)
              3b. 400 Bad Method + SOAP message (Fault result)

              OK transport + Fault SOAP
              4. 200 OK + SOAP message (Fault result)

              > In that case, what should I get as the second parameter to the
              > fault
              > handler?  In the debugger, @_ has length 2, and print $_[1], but
              > print ref
              > $_[1] returns an empty string.  Is that because the content of the
              > response is empty?
              Exactly. ref $_[1] will return empty string even if $_[1] is not
              empty, but doesn't represent any referense and contains plain string,
              like 'abc'. That means that content of message either empty or not
              parsed.

              Best wishes, Paul.


              __________________________________________________
              Do You Yahoo!?
              Get personalized email addresses from Yahoo! Mail - only $35
              a year!  http://personal.mail.yahoo.com/


              To unsubscribe from this group, send an email to:
              soaplite-unsubscribe@yahoogroups.com


            Your message has been successfully submitted and would be delivered to recipients shortly.