Re: [Openoss-devel] SOAP Gen status

[Flauw, M. <Mar...@hp...>]

Dear all,

Most of the problems we are finding now are due to the fact we don't have good regression tests in place and nobody to maintain and run them.
A number of those regression tests were developed and run on JOSIF V1.0 by Jin-Zhai Zhang along with the development of the initial specification generators (Doc Gen and SOAP Gen), as well as the initial versions of the Model validator, SID Import and Workbench installer. Jin-Zhai worked also full time during 2 years on this project.

We tried to find a resource to maintain and run the regression tests, but even if I spent time coaching, we didn't succeed.  I share also Craig's overall concerns. Xose did some very good work and Kumar is ramping up nicely, but we need more.

Regarding PM catalyst, I am not involved in it, so I cannot tell.  The only thing I can tell is that I was contacted last week by a participant on how to use the interface, showing that at least for this vendor, the work has not started yet.

Best regards

Marc

From: Craig Gallen [mailto:gal...@go...] On Behalf Of Craig Gallen (opennms)
Sent: Thursday, September 27, 2012 7:57 AM
To: Flauw, Marc
Cc: Pierre Gauthier; Xose Ramon Sousa Vazquez (xr...@op...); Tina O Sullivan; SB Mahapatra; openoss-devel; Joann O'Brien; Craig Gallen (OpenNMS)
Subject: Re: SOAP Gen status

Hi Pierre,

Status: SoapImpl, EJBImpl, JVTSpec generators now generating code which compiles for PM, MPAC and RAM agaist the XML generated by the SOAP generator. All ready to package. I will be tagging 1.1.2 and moving to trunk to 1.1.3-SNAPSHOT this week.

I appreciate and share your concerns. Unfortunately all of the problems you mention are well understood. We just don't have the resources with the dedicated time and skills to deal with them. Most of the recent problems we have had have been because nobody was checking the generated XML until I tried to get the implementation code to work. I'm surprised the PM Catalyst folks hadn't noticed the problems with the generated SOAP long before now. This perhaps is a very strong proof of the value of working on RI CTK implementations at the same time as creating the spec.

Personally my biggest problem is that I am not able to put nearly as much time into the work as I did when I was working full time on it during the first two years. If we could find a sponsor, I could put in more effort.

I will be at management world Orlando. It would be great if we could use the opportunity to do a face to face review of where we are and also go through the inventory code. I know you will be teaching some of the time but I will be there from Sunday through Friday morning.

Apart from a technical review, I would also like to talk through what is realistically required to make a sustainable tool chain. Tina and Ken and Joan may also want to join us for some of this. So far we have tried to do this without a realistic appreciation of the costs which have mostly been hidden. There are real costs of sustaining a project which mostly OpenNMS have  funded to get things started (my time and efforts).

I understand the TM Forum paying for you to do  work but I think that is very much future development work and bringing on new participants. It's not the same as the basic job of infrastructure and release management.  Last year the TM Forum did stump up some cash for me to help write marketing collateral and spend the Christmas holidays tagging a release.  Perhaps this is something we could look at on a more sustained basis.

cheers,
Craig



On 27/09/2012 06:13, Flauw, Marc wrote:
Pierre,

I am sorry you were not able to attend the status meeting yesterday. Yes, Xose fixed it on Monday and I confirmed on Tuesday.

Best regards

Marc

From: Pierre Gauthier [mailto:pga...@tm...]
Sent: Wednesday, September 26, 2012 10:49 PM
To: Flauw, Marc; Craig Gallen (opennms)
Cc: Xose Ramon Sousa Vazquez (xr...@op...<mailto:xr...@op...>); Tina O Sullivan; SB Mahapatra
Subject: Re: SOAP Gen status

Hi All,

Have we fixed the issue with the associations in the dependent model.

We should really think about building a regression test suite to handle these
core infrastructural requirements.

We can't wait at the last moment like we do now to figure out so fundamental problems.

This applies to the RI/CTK generation code which has a large number of issues (i.e String Array, Class Name clashes etc...)

Pierre
On Sep 26, 2012, at 3:28 AM, "Flauw, Marc" <Mar...@hp...<mailto:Mar...@hp...>> wrote:



Hi Xose,

Just checking with you.
Did you made any further update to the SOAP Generator since the version I tested yesterday?

Also, did your update the  TIP_SOAP_Generator_Package project with latest jar ?

Best regards

Marc