axis-c-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From John Hawkins <HAWKI...@uk.ibm.com>
Subject Re: Code freeze for Axis C++ 1.3 final release.
Date Thu, 14 Oct 2004 09:53:07 GMT




OK, well that excellent summary tells me that we aren't going to do this
for 1.3 and we should prioritise it for 1.4.

My view - It's too risky, too big  and too late for 1.3.


John Hawkins




                                                                           
             Samisa Abeysinghe                                             
             <samisa_abeysingh                                             
             e@yahoo.com>                                               To 
                                       Apache AXIS C Developers List       
             14/10/2004 10:45          <axis-c-dev@ws.apache.org>          
                                                                        cc 
                                                                           
             Please respond to                                     Subject 
              "Apache AXIS C           Re: Code freeze for Axis C++ 1.3    
             Developers List"          final release.                      
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           




I have been stugling with 136 just to understand what exactly is going
wrong.

I *think* I know how to solve this. Basically there is an error on the way
complex objects are
serialized for RPC style on server side. (Not sure if the same apply for
the doc style)
This would prevent the Axis C++ server from interoperating with other SOAP
engines. (However it
works with Axis C++ clinets) This is *not* a client side problem, and that
is why C++ client can
work with Java server. (and that is one of the reasons why I was not
seriously looking into this
issues as I was working more on the client side)

If we are to fix this we need to change not only the WSDL tool but also the
core serializer code
(namely SoapSerializer::addOutputCmplxParam and/or Param::serialize() and
the generated method for
complex types Axis_Serialize_ComplexTypeName) as per my understanding as of
now. However the
number of methods to be changed would expand as we go on fixing this.

I have been looking into the ways of providing a quick patch, however, it
is not possible to solve
the problem without changing the serializer code.

While it is possible to take some time (2/3 days to 1 week) and fix this,
it is quite risky to
change this at this moment as it is going to affect both the server side
and client side, specilly
a core module - Serializer. (a fix would affect the client side, because at
the mpment the complex
type serialization is done using the same methods for both client and
server. In order to fix this
issue we need to change the way how serialization is done for server.
However at the moment both
the SOAPSerializer and generated client code uses the same syntax/semantics
for serializing, which
is incorrect)

If we are to do 1.3 final in another, say, 2 weeks time, it is not a good
idea to fix this now as
it would take some time before we know for sure the fix works, considering
its complexity.

Thoughts please...

Thanks,
Samisa...

--- Samisa Abeysinghe <samisa_abeysinghe@yahoo.com> wrote:

> 136 - Not fixed. I have requested a WSDL to recreate the problem, but did
not get it. It would
> be
> easier to if the creeater of the issue could provide the WSDL - else it
would take a bit more
> time
> to look for one and fix it. However, I can have a look into this problem.
>
> 75 - This is an enhancement that I requested. To my knowlege, there was
no specific work on
> this.
> We do not have tests in CVS to test againt this - thus we have to build
one - however I am not
> sure how to do it.
>
> Samisa...
>
>
> --- John Hawkins <HAWKINSJ@uk.ibm.com> wrote:
>
> >
> >
> >
> >
> > I look here ->
> >
>
http://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&pid=10461&priorityIds=2&resolutionIds=-1

> >
> >
> > And see that there are still three issues marked as "Critical". It
looks to
> > me as if 136 is critical - any one else got an opinion - has it been
fixed
> > and I missed it?
> >
> > The other two -
> > 89 - currently being dealt with
> > 75 - has this been fixed - do we have a test to ensure the code has not
> > been regressed?
> >
> > Until we're sure that these have been fixed I'm not happy to say that
we
> > are done for 1.3?
> >
> >
> >
> >
> > John Hawkins
> >
> >
> >
> >
> >

> >              damitha kumarage

> >              <damitha@opensour

> >              ce.lk>
To
> >                                        Apache AXIS C Developers List

> >              13/10/2004 12:37          <axis-c-dev@ws.apache.org>

> >
cc
> >

> >              Please respond to
Subject
> >               "Apache AXIS C           Re: Code freeze for Axis C++ 1.3

> >              Developers List"          final release.

> >

> >

> >

> >

> >

> >

> >
> >
> >
> >
> > Hi all,
> > Now since majority of the bugs at jira are resolved and we need to
start
> > working on 1.4 version for new changes how about making the code freeze
> > on Friday 15th Oct 2004. We may have the final release on Tuesday 19th
> > Oct 2004.
> >
> > thanks
> > damitha
> > On Thu, 2004-10-07 at 14:38, Samisa Abeysinghe wrote:
> > > OK. Looks like majority are on the other side.
> > > Lets go ahead and do the release with other transport.
> > >
> > > (I was pushing for axis2 transport because, for me, many problems
went
> > away - even paser related
> > > stuff such as the issue on AxisBench with Xerces)
> > >
> > > Samisa...
> > >
> > > --- John Hawkins <HAWKINSJ@uk.ibm.com> wrote:
> > >
> > > >
> > > >
> > > >
> > > >
> > > > I think the key question is test -
> > > >
> > > > How long would it take to redo the tests on whatever platforms were
> > tested
> > > > in the past wks ? My assumption is it would take that length of
time
> > again?
> > > > Which would be unnaceptable I think.
> > > >
> > > > Our only way of testing the system is by running the fv suite on
all
> > > > platforms. If we can do that then I'm happy to believe that the
> > transport
> > > > is as good as the other one.
> > > >
> > > > We have got a windows and linux build with aut tests running.
> > > >
> > > > NOTE: we use xml4c as the parser not pure xerces.
> > > >
> > > > Andrew P - have we tested the new transport on those platforms. if
so,
> > what
> > > > was the outcome?
> > > >
> > > >
> > > >
> > > >
> > > > John Hawkins
> > > >
> > > >
> > > >
> > > >
> > > >
> >
> > > >              Samisa Abeysinghe
> >
> > > >              <samisa_abeysingh
> >
> > > >              e@yahoo.com>
> > To
> > > >                                        Apache AXIS C Developers
List
> >
> > > >              07/10/2004 06:41          <axis-c-dev@ws.apache.org>
> >
> > > >
> > cc
> > > >
> >
> > > >              Please respond to
> > Subject
> > > >               "Apache AXIS C           Re: Code freeze for Axis C++
1.3
> >
> > > >              Developers List"          final release.
> >
> > > >
> >
> > > >
> >
> > > >
> >
> > > >
> >
> > > >
> >
> > > >
> >
> > > >
> > > >
> > > >
> > > >
> > > > The point is simple. Old transport has known critical bugs (such as
> > message
> > > > size limits and
> > > > terribly slow). New transport has proven to solve the problems and
fast
> > as
> > > > well. Hence it is not a
> > > > question of different transports for 1.3 beta and 1.3 final. Rather
a
> > > > question of fixing known
> > > > bugs in 1.3 beta in 1.3 final.
> > > >
> > > > Is it a good idea to ship a buggy transport with 1.3 fianl, while
we
> > have
> > > > the solution in the CVS
> > > > already?
> > > >
> > > > Samisa...
> > > >
> > > > --- sanjaya singharage <sanjayas@opensource.lk> wrote:
> > > >
> > > > > Is it wise to change the tranport in the middle of the release?
To
> > > > > elaborate, the 1.3 Alpha and Beta will have one transport and the
> > final
> > > > > release will have a different  new transport. Is this recommended
or
> > ok
> > > > > practise for apache projects?
> > > > >
> > > > > sanjaya.
> > > > >
> > > > > ----- Original Message -----
> > > > > From: "Samisa Abeysinghe" <samisa_abeysinghe@yahoo.com>
> > > > > To: "Apache AXIS C Developers List" <axis-c-dev@ws.apache.org>
> > > > > Sent: Wednesday, October 06, 2004 4:56 PM
> > > > > Subject: Re: Code freeze for Axis C++ 1.3 final release.
> > > > >
> > > > >
> > > > > > Are we making axis2 transport and Xerces based paser lib as
> > defaults
> > > > for
> > > > > 1.3 final release?
> > > > > > If yes, we have to change the docs accordingly.
> > > > > > If no, why not ;-)
> > > > > >
> > > > > > Thanks,
> > > > > > Samisa...
>
=== message truncated ===




_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com



Mime
View raw message