You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-dev@axis.apache.org by Eran Chinthaka <ch...@opensource.lk> on 2007/06/03 03:35:25 UTC
Re: Setting must understand property true for all required attributes
in addressing
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hi Amila,
Can you please explain what this issue is? Our current addressing code
passed all the test case during the last interop. IIRC, there were tests
to see whether we understand those required properties (please see
defined set of problems in [1]).
So I assume we are good in terms of the spec. We can not simply change
something because MSFT is doing that or any other company is doing that.
I hate doing so. For example, when we send a request with ReplyTo set to
NONE, one implementation was sending the response using the same channel
that we use to send the request, which is completely wrong. If you want
I can give you some more examples that some of the companies were doing
without adhering to the specs.
So please explain what the problem is and if it is something that we are
doing wrong then let's fix it. You will have to point to the spec and
not to any proprietary implementation please.
Thanks,
Eran Chinthaka
[1] : http://www.w3.org/TR/ws-addr-soap/#soapfaults
Amila Suriarachchi wrote:
> hi,
> Currently, the must understand property of the addressing headers do not
> set for the
> required addressing headers. There is no problem with this, since there
> is no compulsory requirement for that.
> But MS implementation of WS* (WSF) set these attributes.
> Therefore sometimes we get interop issues when processing the must
> understand properties. So setting
> must understand property true within the Axis2 would resolve these
> incompatibility issues easily.
>
> So shall we switch on the must understand property for all the required
> addressing headers if there is no
> reason for not to do so.
> I think it is better to do the same thing for rampart and sandesha too.
>
> if we decide to do so can someone go ahead and do the changes since I am not
> much familiar with the addressing.
>
> thanks,
> Amila.
>
>
> --
> Amila Suriarachchi,
> WSO2 Inc.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGYhrdjON2uBzUhh8RAlwpAJ4zToh2QACxCRoT/ByHVIwvO6rkmACfT65u
WDsFB1pmShG43jf2VvIBj7M=
=uGgM
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org
Re: Setting must understand property true for all required attributes in addressing
Posted by David Illsley <da...@gmail.com>.
:-) Yes I will, though I want to see the spec progress a bit before I
tidy up what I've got and check it in.
David
On 04/06/07, Eran Chinthaka <ch...@opensource.lk> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hey,
>
> I am sorry If my wording caused you some problem, I didn't mean it :). I
> am bit worried if some one says company X does so, so we also should do
> that, without adhering to the spec.
>
> > We can not simply change
> > something because MSFT is doing that or any other company is doing
> > that.
> > I hate doing so.
> >
> >
> >> 100% agree. I propose this not because MSFT is doing this. but I saw
> >> some advantage of doing this in terms of fixing our bugs early as
> >> possible (once again this does not mean addressing has bugs).
> >> That is why is I think davids' solution is fair enough.
>
> Totally agreed. If there is a bug let's discuss it. Can you please put
> down the problem and the solution so that we can agree for the final
> decision. These days there are too much of discussion happening parallel :)
>
> BTW, it will be great if we can implement WS-Addressing policies. IIRC,
> David promised me he will do it, when we met last time ;). We have the
> required functionality, so David will glue them together.
>
> Chinthaka
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.3 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFGY3ufjON2uBzUhh8RAgO+AJ96Fx3g/Lxx2bjp/D62THJ3QfRiDACcD0V+
> BlzZUyNyByJ7gbxzRcnafsY=
> =svC6
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> For additional commands, e-mail: axis-dev-help@ws.apache.org
>
>
--
David Illsley - IBM Web Services Development
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org
Re: Setting must understand property true for all required attributes
in addressing
Posted by Eran Chinthaka <ch...@opensource.lk>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hey,
I am sorry If my wording caused you some problem, I didn't mean it :). I
am bit worried if some one says company X does so, so we also should do
that, without adhering to the spec.
> We can not simply change
> something because MSFT is doing that or any other company is doing
> that.
> I hate doing so.
>
>
>> 100% agree. I propose this not because MSFT is doing this. but I saw
>> some advantage of doing this in terms of fixing our bugs early as
>> possible (once again this does not mean addressing has bugs).
>> That is why is I think davids' solution is fair enough.
Totally agreed. If there is a bug let's discuss it. Can you please put
down the problem and the solution so that we can agree for the final
decision. These days there are too much of discussion happening parallel :)
BTW, it will be great if we can implement WS-Addressing policies. IIRC,
David promised me he will do it, when we met last time ;). We have the
required functionality, so David will glue them together.
Chinthaka
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGY3ufjON2uBzUhh8RAgO+AJ96Fx3g/Lxx2bjp/D62THJ3QfRiDACcD0V+
BlzZUyNyByJ7gbxzRcnafsY=
=svC6
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org
Re: Setting must understand property true for all required attributes in addressing
Posted by Amila Suriarachchi <am...@gmail.com>.
On 6/3/07, Eran Chinthaka <ch...@opensource.lk> wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Amila,
>
> Can you please explain what this issue is? Our current addressing code
> passed all the test case during the last interop. IIRC, there were tests
> to see whether we understand those required properties (please see
> defined set of problems in [1]).
First of all I should apologise you If I mean any thing bad about
addressing. Sorry I really can't
remember the case and if so I would have log a jira. may be my configuration
problem.
Actually In the first mail I wanted to write "some times we might get " but
I had written it as
"some times we get" (you know my writing, once again sorry for that.) and
propose this (which came to my mind at that time) as a precautionary step.
So I assume we are good in terms of the spec.
yes I think as well.
We can not simply change
> something because MSFT is doing that or any other company is doing that.
> I hate doing so.
100% agree. I propose this not because MSFT is doing this. but I saw some
advantage of doing this in terms of fixing our bugs early as possible (once
again this does not mean addressing has bugs).
That is why is I think davids' solution is fair enough.
For example, when we send a request with ReplyTo set to
> NONE, one implementation was sending the response using the same channel
> that we use to send the request, which is completely wrong.If you want
>
I can give you some more examples that some of the companies were doing
> without adhering to the specs.
>
> So please explain what the problem is and if it is something that we are
> doing wrong then let's fix it. You will have to point to the spec and
> not to any proprietary implementation please.
>
> Thanks,
> Eran Chinthaka
>
> [1] : http://www.w3.org/TR/ws-addr-soap/#soapfaults
>
> Amila Suriarachchi wrote:
> > hi,
> > Currently, the must understand property of the addressing headers do not
> > set for the
> > required addressing headers. There is no problem with this, since there
> > is no compulsory requirement for that.
> > But MS implementation of WS* (WSF) set these attributes.
> > Therefore sometimes we get interop issues when processing the must
> > understand properties. So setting
> > must understand property true within the Axis2 would resolve these
> > incompatibility issues easily.
> >
> > So shall we switch on the must understand property for all the required
> > addressing headers if there is no
> > reason for not to do so.
> > I think it is better to do the same thing for rampart and sandesha too.
> >
> > if we decide to do so can someone go ahead and do the changes since I am
> not
> > much familiar with the addressing.
> >
> > thanks,
> > Amila.
> >
> >
> > --
> > Amila Suriarachchi,
> > WSO2 Inc.
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.3 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFGYhrdjON2uBzUhh8RAlwpAJ4zToh2QACxCRoT/ByHVIwvO6rkmACfT65u
> WDsFB1pmShG43jf2VvIBj7M=
> =uGgM
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> For additional commands, e-mail: axis-dev-help@ws.apache.org
>
>
--
Amila Suriarachchi,
WSO2 Inc.
Re: Setting must understand property true for all required attributes
in addressing
Posted by Eran Chinthaka <ch...@opensource.lk>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Amila Suriarachchi wrote:
> I am sorry I can not understand what you are trying to explain here. Can
> you please elaborate a bit?
>
>
>> Lets take this senario of a inout message.
>
>> First client send the request to server. since we do not have set the
>> mustunderstand true, the addressing headers does not contain it. so if
>> we want to have it we can set it using the client options.
>
>> Then server gets the request and send a reply back to client. for this
>> request we can not set the
>> must understand properties true if it is not the default may. I don't
>> know whether it is possible using the axis2.xml or the services.xml.
Amila, sorry If I forgot something. Why do you wanna set mustUnderstand
true explicitly for addressing headers? With the earlier fix won't it
fix this too?
Even if you do that, then I think the problem is somewhere else.
Processing addressing headers is Addresssing module's problem. Why do
you wanna mess with it?
Chinthaka
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGZQJ5jON2uBzUhh8RAiZ6AKCgIA/Bu8eaOguoaHkyupeYZZi2rACeKmtj
oyHU9vdMTH4fprHZ/UB49LI=
=dREi
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org
Re: Setting must understand property true for all required attributes in addressing
Posted by Amila Suriarachchi <am...@gmail.com>.
On 6/5/07, Eran Chinthaka <ch...@opensource.lk> wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Amila Suriarachchi wrote:
> > I have fixed this by changing the phases. i.e. security come before
> > addressing.
>
> I don't think changing phases is the solution for this.
yes correct solution is to put the addressing before security and fix the
rampart issue.
If some one set
> isProcessed as true, then no-one better not set it to false, unless
> there is a very special reason.
>
> Why I put that flag is to make must understand checker to know which
> headers are processed and which ones are not. So all the module authors
> should adhere to it. Rather than changing the phases. can u please make
> sure rampart handlers does the proper thing?
yes, I informed ruchith and now he should have fixed it. What I am telling
is if we had
set mustunderstand true by default some one would have seen this much
earlier.
>
> > When cosidering this senario I see some advantage of setting this flag.
> > Although there is no problem with
> > the addressing. We can swith on the mustunderstand
> > in client side using client options but can not think a way in server
> > side. (when sending the request)
>
> I am sorry I can not understand what you are trying to explain here. Can
> you please elaborate a bit?
Lets take this senario of a inout message.
First client send the request to server. since we do not have set the
mustunderstand true, the addressing headers does not contain it. so if we
want to have it we can set it using the client options.
Then server gets the request and send a reply back to client. for this
request we can not set the
must understand properties true if it is not the default may. I don't know
whether it is possible using the axis2.xml or the services.xml.
Amila.
Thanks,
> Chinthaka
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.3 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFGZHjGjON2uBzUhh8RAu/pAJ9QJ6cit4oehaKWSoueIOiS47dxiwCeK5iK
> KineLK0LfqI+McNtgD/R3tI=
> =PAZN
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> For additional commands, e-mail: axis-dev-help@ws.apache.org
>
>
--
Amila Suriarachchi,
WSO2 Inc.
Re: Setting must understand property true for all required attributes
in addressing
Posted by Eran Chinthaka <ch...@opensource.lk>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Amila Suriarachchi wrote:
> I have fixed this by changing the phases. i.e. security come before
> addressing.
I don't think changing phases is the solution for this. If some one set
isProcessed as true, then no-one better not set it to false, unless
there is a very special reason.
Why I put that flag is to make must understand checker to know which
headers are processed and which ones are not. So all the module authors
should adhere to it. Rather than changing the phases. can u please make
sure rampart handlers does the proper thing?
>
> When cosidering this senario I see some advantage of setting this flag.
> Although there is no problem with
> the addressing. We can swith on the mustunderstand
> in client side using client options but can not think a way in server
> side. (when sending the request)
I am sorry I can not understand what you are trying to explain here. Can
you please elaborate a bit?
Thanks,
Chinthaka
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGZHjGjON2uBzUhh8RAu/pAJ9QJ6cit4oehaKWSoueIOiS47dxiwCeK5iK
KineLK0LfqI+McNtgD/R3tI=
=PAZN
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org
Re: Setting must understand property true for all required attributes in addressing
Posted by Amila Suriarachchi <am...@gmail.com>.
finally I found the bug. This is the reason.
In rampart before doing any processing it creates an envelop of the original
soap envelop.
in this process it does not copy the isProccessed attributes set in the
SoapHeaderBlock. As a result of this
if the isProcessed atribute is set then after this process again it set to
false.
in the AddressingInhadlers it process Header Blocks and set the attributes.
So if we have Security after the addressing, those is processed attributed
set by the AddressingInHadlers is set to false and this gives an exception
at the checkMustunderstand method.
I have fixed this by changing the phases. i.e. security come before
addressing.
When cosidering this senario I see some advantage of setting this flag.
Although there is no problem with
the addressing. We can swith on the mustunderstand
in client side using client options but can not think a way in server side.
(when sending the request)
Amila.
On 6/4/07, Jaliya Ekanayake <jn...@gmail.com> wrote:
>
> Hi Amila and Eran,
>
> I also don't think that axis2 should set mustunderstand to all the headers
> unless it is required by the application.
> For e.g. say a client needs to inform the server that it should understand
> the <wsa:Action> by setting mustunderstand =1 for wsa:Action.
> So if the server does not process the wsa:Action a fault should be
> generated.
>
> From the receiving side, each module implementation should check the
> mustunderstand property and "understand" it(we have this right now) but I
> am
> not sure whether we can set the mustunderstand property for specific SOAP
> headers by the axis2 client.
>
> Thanks,
> -jaliya
>
>
>
>
> ----- Original Message -----
> From: "Eran Chinthaka" <ch...@opensource.lk>
> To: <ax...@ws.apache.org>
> Sent: Saturday, June 02, 2007 9:35 PM
> Subject: Re: Setting must understand property true for all required
> attributes in addressing
>
>
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > Hi Amila,
> >
> > Can you please explain what this issue is? Our current addressing code
> > passed all the test case during the last interop. IIRC, there were tests
> > to see whether we understand those required properties (please see
> > defined set of problems in [1]).
> >
> > So I assume we are good in terms of the spec. We can not simply change
> > something because MSFT is doing that or any other company is doing that.
> > I hate doing so. For example, when we send a request with ReplyTo set to
> > NONE, one implementation was sending the response using the same channel
> > that we use to send the request, which is completely wrong. If you want
> > I can give you some more examples that some of the companies were doing
> > without adhering to the specs.
> >
> > So please explain what the problem is and if it is something that we are
> > doing wrong then let's fix it. You will have to point to the spec and
> > not to any proprietary implementation please.
> >
> > Thanks,
> > Eran Chinthaka
> >
> > [1] : http://www.w3.org/TR/ws-addr-soap/#soapfaults
> >
> > Amila Suriarachchi wrote:
> >> hi,
> >> Currently, the must understand property of the addressing headers do
> not
> >> set for the
> >> required addressing headers. There is no problem with this, since there
> >> is no compulsory requirement for that.
> >> But MS implementation of WS* (WSF) set these attributes.
> >> Therefore sometimes we get interop issues when processing the must
> >> understand properties. So setting
> >> must understand property true within the Axis2 would resolve these
> >> incompatibility issues easily.
> >>
> >> So shall we switch on the must understand property for all the required
> >> addressing headers if there is no
> >> reason for not to do so.
> >> I think it is better to do the same thing for rampart and sandesha too.
> >>
> >> if we decide to do so can someone go ahead and do the changes since I
> am
> >> not
> >> much familiar with the addressing.
> >>
> >> thanks,
> >> Amila.
> >>
> >>
> >> --
> >> Amila Suriarachchi,
> >> WSO2 Inc.
> >
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.4.3 (GNU/Linux)
> > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> >
> > iD8DBQFGYhrdjON2uBzUhh8RAlwpAJ4zToh2QACxCRoT/ByHVIwvO6rkmACfT65u
> > WDsFB1pmShG43jf2VvIBj7M=
> > =uGgM
> > -----END PGP SIGNATURE-----
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> > For additional commands, e-mail: axis-dev-help@ws.apache.org
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> For additional commands, e-mail: axis-dev-help@ws.apache.org
>
>
--
Amila Suriarachchi,
WSO2 Inc.
Re: Setting must understand property true for all required attributes in addressing
Posted by Jaliya Ekanayake <jn...@gmail.com>.
Hi Amila and Eran,
I also don't think that axis2 should set mustunderstand to all the headers
unless it is required by the application.
For e.g. say a client needs to inform the server that it should understand
the <wsa:Action> by setting mustunderstand =1 for wsa:Action.
So if the server does not process the wsa:Action a fault should be
generated.
>From the receiving side, each module implementation should check the
mustunderstand property and "understand" it(we have this right now) but I am
not sure whether we can set the mustunderstand property for specific SOAP
headers by the axis2 client.
Thanks,
-jaliya
----- Original Message -----
From: "Eran Chinthaka" <ch...@opensource.lk>
To: <ax...@ws.apache.org>
Sent: Saturday, June 02, 2007 9:35 PM
Subject: Re: Setting must understand property true for all required
attributes in addressing
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Amila,
>
> Can you please explain what this issue is? Our current addressing code
> passed all the test case during the last interop. IIRC, there were tests
> to see whether we understand those required properties (please see
> defined set of problems in [1]).
>
> So I assume we are good in terms of the spec. We can not simply change
> something because MSFT is doing that or any other company is doing that.
> I hate doing so. For example, when we send a request with ReplyTo set to
> NONE, one implementation was sending the response using the same channel
> that we use to send the request, which is completely wrong. If you want
> I can give you some more examples that some of the companies were doing
> without adhering to the specs.
>
> So please explain what the problem is and if it is something that we are
> doing wrong then let's fix it. You will have to point to the spec and
> not to any proprietary implementation please.
>
> Thanks,
> Eran Chinthaka
>
> [1] : http://www.w3.org/TR/ws-addr-soap/#soapfaults
>
> Amila Suriarachchi wrote:
>> hi,
>> Currently, the must understand property of the addressing headers do not
>> set for the
>> required addressing headers. There is no problem with this, since there
>> is no compulsory requirement for that.
>> But MS implementation of WS* (WSF) set these attributes.
>> Therefore sometimes we get interop issues when processing the must
>> understand properties. So setting
>> must understand property true within the Axis2 would resolve these
>> incompatibility issues easily.
>>
>> So shall we switch on the must understand property for all the required
>> addressing headers if there is no
>> reason for not to do so.
>> I think it is better to do the same thing for rampart and sandesha too.
>>
>> if we decide to do so can someone go ahead and do the changes since I am
>> not
>> much familiar with the addressing.
>>
>> thanks,
>> Amila.
>>
>>
>> --
>> Amila Suriarachchi,
>> WSO2 Inc.
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.3 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFGYhrdjON2uBzUhh8RAlwpAJ4zToh2QACxCRoT/ByHVIwvO6rkmACfT65u
> WDsFB1pmShG43jf2VvIBj7M=
> =uGgM
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
> For additional commands, e-mail: axis-dev-help@ws.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: axis-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: axis-dev-help@ws.apache.org