|
I have
(finally) been able to get some of the auto test
cases running.
The
most important hint came from Roshan discouraging the use of
SimpleAxisServer.
This
looks like we need to point this out further. Using the TCPMonitor from the test
suite together with
the
SAS gave me always a Connection reset by peer 10054. Has somebody been able
to
use
SAS with the test cases?
After
switching to apache things started working. Since I need to get the services in
place
- we
are missing so far an automatic build on the server side - I could so far only
construct 5 test cases:
AxisBench, CalculatorDoc, FourLevelTestDoc, MathOpsDoc,
SimpleArrayDoc
which
passed using the attached code changes.
I will
need a bit more time to construct the maybe 10 other relevant test cases, so if
John wants to try
it
that would be great. But please with the attached diffs : one replaces the
cpp\literal\BeanParamWriter.java
from
the message below and the other one is a change in the array section of the
src\soap\SoapDeSerializer.cpp
which could use some review in any case and the
cpp\literal\ClientStubWriter.cpp from below.
The
other option is to trust the so far committed serializer changes (with more
confidence now) and build
the
alpha and the attached changes will be committed after alpha with add.
testing.
Thanks,
Carsten
Hi john,
could you confirm whether you can test the
attached changes by carsten?
sanjaya.
----- Original Message -----
Sent: Tuesday, February 01, 2005 12:45
AM
Subject: RE: Axis c++ 1.5 release plan
(draft 4)
Sorry for the late reply but my timezone (PST) seems to be a bit
different than most other people here.
First of all, the changes are only related to the document
literal serialization and code generation
in
cpp mode. RPC style and c stubs are
unaffected.
I
have had trouble running the test suite in tests/auto_build (with the two
services I have there, it seems
the request doesn't make it to the service), but I have run the
relevant examples in samples (doclitbase,
base,doclitgroupB,groupB) successfully, which include the mesa
interoperability tests.
This is insufficient for testing the code generation changes (since
the stub are statically checked in) but it
does test the serializer changes - that's where I got some confidence
that the serializer changes
won't wreak havoc.
More details in case somebody wants to run the
changes:
The diffs for the two files representing the code generation changes
are attached.
These are
src/wsdl/org/apache/axis/wsdl/wsdl2ws/cpp/literal/BeanParamWriter.java
and
src/wsdl/org/apache/axis/wsdl/wsdl2ws/cpp/literal/ClientStubWriter.java
I
also attached the test case I intend to add into tests/auto_build (The
FourLevelTestDoc.wsdl into
tests/auto_build/testcases/wsdl, the
FourLevelTestDoc.xml file into
tests/auto_build/testcases/wsdl,
the test client FourLevelTestDocClient.cpp into
tests/auto_build/testcases/client/cpp,
the test server FourLevelTestDocInterface.cpp into
tests/auto_build/testcases/client/cpp
and the
expected output FourLevelTestDoc.expected into
tests/auto_build/testcases/output).
I
will keep plugging away at this to run these on my own. BTW, so far I have
trying to use the
SimpleAxisServer - would it be better to use the apache
plugin?
Thanks,
Carsten
No problem. as long as we get unconfused :).
(I think there was a full stop missing in this reply and I have put it
there to avoid further confusion :) )
Carsten, could you take note and respond
please?
sanjaya.
----- Original Message -----
Sent: Monday, January 31, 2005 7:31
PM
Subject: Re: Axis c++ 1.5 release
plan (draft 4)
hi,
sorry for the confusion :-)
I suggest we ask Carsten to run the regression
tests to ensure that his code gen changes have not broken stuff on
at least one platform of his choice. He can then decide, based on
this test results, which he can discuss here if necessary, whether
he wants to commit his changes or not. If he runs the suite and gets
some errors then I suggest we're happy. if he breaks that suite entirely
then we're not?
Running the
suite should take him three hours from scratch, if he hasn't done it
before so we should know soon whether the gen changes are viable.
If he is not able to test them (not
having the services would be the obvious issue) then perhaps one of us
can run them for him (Carsten - have you got a branch we can
extract?).
If Carsten is not
around or able to commit changes then I suggest we go alpha without the
gen changes and do the same as above in alpha timeframe.
John
Hawkins
| "sanjaya
singharage" <[EMAIL PROTECTED]>
31/01/2005 13:09
|
Please
respond to
"Apache AXIS C Developers
List" |
|
|
To
| "Apache AXIS C
Developers List" <[email protected]>
|
|
cc
|
|
|
Subject
| Re: Axis c++ 1.5
release plan (draft 4) |
|
(If the situation is not clear, Carseten has already
put his changes to the serializer. He has also finished the changes to
the code generator but is not able to test them as yet, therefore he has
avoided committing those changes.)
What exactly do you mean? That we go ahead
with the codefreeze with the serilizer changes intact and let Carsten
commit the code generation stuff before the final?
If so we can codefreeze early tuesday morning and
start on the release procedure.
sanjaya.
-----
Original Message -----
From: John
Hawkins
To:
Apache AXIS C Developers List
Sent: Monday, January 31, 2005 6:45
PM
Subject: Re: Axis c++ 1.5 release plan
(draft 4)
I would
suggest Carsten put in his fixes asap (into the Alpha preferably?) but
not hold up alpha or beta?
John Hawkins
| "sanjaya
singharage" <[EMAIL PROTECTED]>
31/01/2005 12:27
|
Please
respond to
"Apache AXIS C Developers
List" |
|
|
To
| "Apache AXIS C Developers List" <[email protected]>
|
|
cc
|
|
|
Subject
| Re: Axis c++ 1.5
release plan (draft 4) |
|
So do we wait for
Carsten to comment the serializer changes?
sanjaya.
----- Original Message -----
From:
John Hawkins
To:
Apache AXIS C Developers List
Sent: Monday, January 31, 2005 4:28 PM
Subject:
Re: Axis c++ 1.5 release plan (draft 4)
Hmm, we're changing the paradigm here - putting in new
code in beta !
Howver, this problem is a bug therefore I go with putting
it in beta. As long as it comes with some good tests that broke before
it was implemented and are fixed by its incorporation.
John
Hawkins
| "sanjaya
singharage" <[EMAIL PROTECTED]>
31/01/2005 08:52
|
Please
respond to
"Apache AXIS C Developers
List" |
|
|
To
| "Apache AXIS C Developers List"
<[email protected]>
|
|
cc
|
|
|
Subject
| Re: Axis c++ 1.5
release plan (draft 4) |
|
I think the serializer
changes for AXISCPP-197 should be commented out
before we go
for the 1.5 codefreeze.
Can we have a quick response from the
community on this?
i.e. Do we keep the serializer changes for
AXISCPP-197 in the codebase for
1.5 alpha or comment it out and
include both serializer and codegeneration
changes for the
final?
please respond quickly guys.
sanjaya.
-----
Original Message -----
From: "Carsten Blecken"
<[EMAIL PROTECTED]>
To: "Apache AXIS C Developers List"
<[email protected]>
Sent: Monday, January 31, 2005 12:49
PM
Subject: RE: Axis c++ 1.5 release plan (draft 4)
>
Quick status update :
>
> I wasn't able to test sufficiently
the code generation changes, so I
> did not commit them (i'm
having problems with the testing framework
> but that's a
different story).
>
> The changes to the serializer have
been committed last week. These
> changes should not break the
existing stubs since they mostly add
> new method overloads with
prefixes for the serializer that miss them.
>
> I will try
to get the code generation changes into the beta time frame
> and
was hoping that it was ok to leave the serializer changes in
there.
>
> Carsten
>
> -----Original
Message-----
> From: Carsten Blecken
> Sent: Thursday,
January 27, 2005 8:46 AM
> To: Apache AXIS C Developers List;
Samisa Abeysinghe
> Subject: RE: Axis c++ 1.5 release plan (draft
4)
>
>
> Sorry for the mess everybody - I assumed
friday to be the last day (my
mistake).
>
> In terms of
things already committed : These are changes to the
serializer
which
> add namespace and/or namespace prefixes to
various methods. They are
overloaded
> methods to keep backward
compatibility with current stubs. So I would say
the
> risk is
not too big. There was also some refactoring being done, but I
did
some
> testing on the changes.
>
> The part
where I need more time are the stub/skeleton generator (which
employ
the
> new methods) and the testing involved with it.
> Maybe
I should hold off the generator changes until after alpha, since
I
consider
> them higher risk?
>
> In terms of
importance: Our current wsdl's break due to the lack
of
namespace
> prefixing. Almost all more complicated doc lit
wsdl's will have issues
here when
> they have data type nesting
and arrays.
>
> Carsten
>
> -----Original
Message-----
> From: Samisa Abeysinghe
[mailto:[EMAIL PROTECTED]
> Sent: Thursday, January 27,
2005 3:37 AM
> To: Apache AXIS C Developers List
> Subject:
Re: Axis c++ 1.5 release plan (draft 4)
>
>
> Part of
the solution has been already commited.
> So we have to either
hold off or comment out what is already commited.
>
> How
critical/important is this feature?
>
> Is it practicle to
tag the CVS and start on testing cycle till the
> rest of the
solution is commited.
> Given that the solution is related to
serializer, the risk of side
> effects is high.
>
>
Thanks,
> Samisa...
>
>
> On Thu, 27 Jan 2005
09:16:35 +0000, John Hawkins
<[EMAIL PROTECTED]>
wrote:
> >
> >
Hmm,
> >
> > Should we hold off on alpha until it's
complete?
> >
> >
> >
> >
>
> John Hawkins
> >
> >
> >
>
>
> > "Carsten Blecken"
<[EMAIL PROTECTED]>
> >
> > 27/01/2005
08:39
> > Please respond to
> > "Apache AXIS C
Developers List"
> > To"Apache AXIS C Developers List"
<[email protected]>,
> >
<[EMAIL PROTECTED]>
> > cc
> > SubjectRE:
Axis c++ 1.5 release plan (draft 4)
> >
> >
>
>
> >
> >
> >
> >
>
>
> > Hi Sanjaya,
> >
> > still working on
item 7). Harder than I thought. Partially committed
> >
(Serializer changes) and I think I can get the rest
> > (code
generator) by friday into the alpha.
> >
> > Do you
want status by friday again?
> >
> > Carsten
>
>
> > -----Original Message-----
> > From: sanjaya
singharage [mailto:[EMAIL PROTECTED]
> > Sent: Wednesday,
January 26, 2005 10:30 PM
> > To: Apache AXIS C Developers
List
> > Subject: Axis c++ 1.5 release plan (draft 4)
>
>
> > Hi all,
> >
> > COULD I HAVE THE
STATUS OF ITEM NUMBER 7 CARSTEN, IS IT ALREAD COMMITED
IN
>
> CVS? CAN WE INCLUDE IT IN 1.5 OR SHELF IT FOR THE NEXT RELEASE? OR
ARE
YOU
> > PLANNING TO FIX THIS BY 1.5 FINAL? COULD I HAVE
A RESPONSE CARSTEN?
> >
> >
> > MARK, WHAT IS
THE STATUS OF ITEM 2? COULD YOU ALSO RESPOND?
> >
>
>
> > Following the discussion I have added these
items
> > -Problem with complex types which contain complex
types
> > http://issues.apache.org/jira/browse/AXISCPP-197 (in
progress by
carsten)
> >
> > Following the
discussion I have removed the following items
> >
> >
-Fault mapping (stub level)
> > -improvement of Web
Site
> > a.Web-site should be made more or less like Axis-Java
website for ease
of
> > use
> > b.More user
oriented and less developer oriented
> > -Add support for
xsd:any type (?)
> > -Ability to add custom headers (?)
>
>
> >
> > sanjaya.
> >
> >
=======================================================
>
>
> > (fourth draft of the release plan.)
>
>
> > Hi all,
> > I would like to volunteer to be
the release manager for the Axis C++ 1.5
> > release.
>
>
> > A draft Release Plan is presented below using the
current items found in
the
> > wiki at
http://wiki.apache.org/ws/FrontPage/AxisC_2b_2b/ToDo1.5 .
> >
Please update the wiki of any more items that are applicable to the
1.5
> > release that developers are working on.
>
>
> > Also we need to decide on dates for code freeze etc.
and release dates
at
> > the bottom.
> > Please
comment.
> >
> >
> >
> > This is the
release plan (draft) for Axis C++ version 1.5. Given below
is
a
> > snapshot
> > of all the changes for 1.5 since
the 1.4 version.
> >
> > Changes from 1.4 to
1.5:
> >
> > 1. Partial C
support -
> > http://issues.apache.org/jira/browse/AXISCPP-210
(in progress by Mark)
> > -C clients with stub generation for
simple types
> > -Porting of some C++ tests to C
> >
2. Memory clean-up and fix -
> >
http://issues.apache.org/jira/browse/AXISCPP-149 (in progress by
Mark
> > Whitlock as per JIRA)
> > 3.
Improve HTTP Support -
> >
http://issues.apache.org/jira/browse/AXISCPP-176 (Done by Samisa as
per
> > JIRA)
> > 4. "100
Continue" support in axis2 transport
> >
http://issues.apache.org/jira/browse/AXISCPP-311 (Done by
Samisa)
> > 5. Proxy support for
axis2 transport
> >
http://issues.apache.org/jira/browse/AXISCPP-325 (Done by
Samisa)
> > 6. Numerous memory leak
fixes (Done by Samisa)
> > 7.
Problem with complex types which contain complex types
>
> http://issues.apache.org/jira/browse/AXISCPP-197 (in progress
by
Carsten)
> >
> > Details on these features could
be found on JIRA and the AxisCpp TODO
wiki
> > page at
http://wiki.apache.org/ws/FrontPage/AxisC_2b_2b/ToDo1.5
>
>
> > Release Criteria:
> >
> > -
Functional tests pass
> > - Test on installation
procedures
> > - Test on different versions of Apache
web server(1.3.xx, 2.0.xx)
> > - Test samples on
different platforms (Linux, Microsoft Windows)
> > - Fix
blockers and most critical bugs that applies to 1.5
>
>
> > Dates:
> >
> > The following are the
proposed dates for the release, opinions please:
> >
Code freeze
January 27, 2005
> > Release alpha
version January 31, 2005
> >
Release 1.5 final
February 21, 2005
> >
> >
> >
sanjaya.
> >
>
>
>
|
BeanParamWriter.java.diff