Brian Smith

Ranch Hand
+ Follow
since Oct 26, 2002
Merit badge: grant badges
For More
Cows and Likes
Cows
Total received
In last 30 days
0
Forums and Threads

Recent posts by Brian Smith

You're awesome, Ruijin ! This is what I was looking for. Thanks for taking your time to explain these steps. I wanted to reinforce my understanding that HelloIF is generated by wscompile at compile time!

Originally posted by ruijin yang:

dynamicproxy.HelloIF myProxy = (dynamicproxy.HelloIF)helloService.getPort(
new QName(nameSpaceUri,portName),
dynamicproxy.HelloIF.class);
1) HelloIF.class is generated by wscompile (at compile time)
2) The port name (HelloIFPort) is specified by the WSDL file
3. Compile Client Code with Service Endpoint Interface class CLASSPATH
4. Package Client to a jar file with SEI class.



I appreciate your help!
Thank you,
I completely understand all the theoritical aspects that you guys are saying here. Now here's what I am asking about. The following codes are taken from Mikalai Zaikin's notes which is a great notes.

This is an example of a Dynamic Proxy Client


Now look at this code line in bold. In order for this code to compile, there must exist HelloIF . My question is, how do we get this? do we generate it from WSDS at compile time as we NEED this for this code to compile. We can't wait until run time? Am I making sense here or I am missing something?
Thanks,
BR
All I am asking is how do you get the HelloIF for a client to use? do you get it generated by webservice tools using WSDL?
Thanks.
Hello MZ and other folks,
I have a question regarding dynamic proxy client. I have gotten stuck at the line where it says the following: (this line is copied from MZ's note)-


How do you get an access to HelloIF? Without an access to it, the code can't even compile. I know we can generate this stub from WSDL, but wouldn't that be similar to stub-based client? It would great if someone can explain to me this concept of getting access to HelloIF dynamically?

Thanks,
B
Hi Durgaparasad,

Originally posted by Durgaprasad Guduguntla:
If we assume that the web container and EJB container are colocated within the same Application server, we can use local interface for the web application and remote interface for the swing client, for accessing the facade to businnes tier. That way we can confine RMI communication to swing client only and on the other hand as the web client uses local communication, there will be a significant improvement in performance. What do you think?



That's what I am considering doing. I am wondering if we can know whether web clients performs better over the swing client or vice-versa. One the requirements is to provide FBN reps faster performing client to support their customers over the phone. Please let me know what you guys think.

BR
Hello Dan,

Originally posted by Dan Drillich:
Hi Brian,



What do you think?

Regards,
Dan



Yes, that's what the web server is meant - serving http requests. Web Server mostly,at least in java world, is like Apache where all the static stuffs like html, images and other css are deployed. Anything to do with dynamic request are routed to Application server where Servlet, JSP, EJB and other business objects are deployed to support business logics. I think the hardwares mentioned in the conversation with CIO can be utilized just this way.
What do you think?
BR
Hello Steve,

Originally posted by Steve Taiwan:
Dear Andrew.

Yes, I designed to do the second query based on the first query. I think this is much efficient.

If there are 10 itineraries from the frsit query and then user selects one, I only have to do the second query once based on the selected itinerary.If I don't do this, I might have 10*10 query in the first query and have to caache them in the Context Object, which is so fat.

I think i won't put a lot of imformation in Context Object and I don't plan to make Context Object a shopping cart. Airline seat booking is different from buying books in Amazon. So I don't make a shooping cart in the design. At the current stage, I have no plan to implement ValueListHandler

What do you think??

[ November 22, 2006: Message edited by: Steve Taiwan ]



I agree with you on the Shopping Cart concept. Its not like other online retailer like Amazon.com, ebay.com etc where you would go and add product on the shopping cart. I am not desiging a shopping cart concept too. However, as far as the Context object vs ValueListHandeler for caching search results, I would prefer ValueListHandler because it lives in EJB tier and thus caching the search results in it would be accessible for all the clients. The Context Object lives in presentation tier plus its design is not intended for caching stuffs particularly the search results.

Also, Guys, FlyByNight already has one Application Server and 2 Web servers, is the intention to separately deploy the EJB plus business objects in Application server and the web tier (Servlets, JSPs, HTML and other stuffs) in those 2-web servers?

Thanks,
BR
[ November 22, 2006: Message edited by: Brian Smith ]
Hello All,
I have been trying to find some documentation that would look into the peformance of Swing Application vs Web Application with no success.can Having Swing Application talk to EJB tier directly improve the speed? Can anyone explain me if one has a better performance (speed) over the other and why?

Thanks,
BR
Andrew,

Originally posted by Andrew Zilahi:


BD is a POJO on the client/web tier, not a SLSB.



Yes according to this CoreJ2EE Patterns, it is a POJO. It also says that BD belongs to Business Tier which will be instantiated by FrontController or Application Client controller to access Service Locator to access session facade to get the business logic execute. Please let me know If I am wrong.

Thanks,
BR
Hello Guys,
This is very interesting dicussion topic and also this area is where I am struggling at. Thanks for initiating this topic.

Originally posted by Marta De rossi:
Hi all !
In fact I don't like very much to use both a Stateful SB and HTTPSession to store the handle(of SFSB). It seems more complicated and error-prone to keep up with both this session information.

Now I re-engineered it with just SLSB. I have inserted the ShoppingCart object in the HTTPSession for web clients and just kept it in memory for plain Java Clients.

I think it could make also the application more scalable.
I hope that my approach will not be a failure when I deliver my assignment


What do you say ?
Thanks
Marta



As far as whether it's gonna work, it absolutely works. However, is this an ideal approach, I am not sure as the sun blue prints, its advisable to store the session in the EJB tier if the EJB tier is presence in the design in such a way that you don't have to maintain it in two places separately for Web Client and Java Client. Also, Is it scalable? I don't know - please how do you think it would be a scalable approach.

If I understood correctly, this dicussion seems to have suggested that it's a good idea to store the session centrally in a Business Delegate what would work both for Web Client and Java Client - leads manageability definitely. But I stil don't know how its gonna be implemented though. Do we store the Session Facade (SFSB) in some sort of collection in the Business Delegate? How do we design the Business Delegate? as a SLSB? How does clients access this BD? through Service Locator? How does the individual client keep its handle to the Session Facade? Please help understand this.

Thanks.
BR
Hello Changski,
Can you share you approach on Business Delegate design? did you create a single Business Delegate? Also, was it a SLSB? How did the clients of it use it?

Thanks,
BR
Hello,
What's your opinion here:

1. Should Class Diagram have ONLY business domain classes or other classes like Business Delegate, Business Facade, ServiceLocater, other helper classes etc?

2. What level of details should the classes have as far as attributes and methods are concerned? should datatype of attributes be shown? should signature of the methods be shown? I know, the Exam instruction does say that we do not have to show everything. but should we not show at least all the methods that are used in sequence diagram to realize the use case?

Thanks,
BR
Congratulation Deepak, Good Sore!
can you tell me what level of details did you have in your class diagram? did you have all the possible methods a class can have or at least all the methods that used in Sequence diagram?

Also, did your methods have their signature?

Thanks.
BR
Congratulation Chris.

Can you please help me with this?UML and Component Diagram

Thanks.
BR
Hello Changski,

Originally posted by Changski Tie Zheng Zhang:
Hi Brian,
I didn't use anything fancy (lollipop and socket), just basic features in Rational Rose will be enough, I think.



Ok, so if I don't use the lollipop and the socket, the remaining notation for component diagram are the dashed arrow and the solid arrow which I believe have the same semantic as in Class diagram ie, dependency and association. Nnow when do I use association vs dependency in Component Diagram?

Also, do you use a bi-directional association?

BR
[ November 08, 2006: Message edited by: Brian Smith ]