John Pritchard

Ranch Hand
+ Follow
since Nov 15, 2002
Merit badge: grant badges
For More
Cows and Likes
Cows
Total received
0
In last 30 days
0
Total given
0
Likes
Total received
0
Received in last 30 days
0
Total given
0
Given in last 30 days
0
Forums and Threads
Scavenger Hunt
expand Ranch Hand Scavenger Hunt
expand Greenhorn Scavenger Hunt

Recent posts by John Pritchard

Thanks for the replies Chris and Kyle. I essentially have a vey large pub/sub architecture where various events are being published to a set of JMS Topics. Our clients would like to pull specific information (messages) off of the topics by using some filter criteria they specify at subscription time. I'm trying to define a mechanism where I can take a client's subscription request and filter criteria and have the filtering occur on the server side to minimize traffic back to the clients. Server side componenets would then asycnronously update the clients when a message did meet their filter criteria.
Appreciate the help.
I know the JMS spec does not mandate it, but does WebSphere's JMS implemenmtation have a Topic Hiereachy implementation (category.subcategory.subcategory) whereby a subscriber could specify wildcards in the subscription request, e.g. subscribe to category.subcategory.*
20 years ago
Hello All-
I've got a system I am designing where a set of several JMS topics exist. Clients would like to subscribe to one or more of these topics anbd provide some filtering criteria as part of the subscription request. The MDB Message Selector semed like an alternative but it doesn't seem that it is dynamically configurable, e.g. specifying the selector at runtime. Any suggestions on a design solution that helps minimize traffic back to the client?
Thanks,
john
Well I received that dreaded email yesterday " We have a new integration project we need you to look at, it has a few legacy systems" !!
Here's the layout. I have 3 C++ applications (2 Wintel and 1 Linux) that essentially listen on a respective port, pick off a message, and plot the information on a map. Each of these systems was developed without knowledge of the other so there's no current data exchange.
I have a fourth system that has some similar funcationality (plotting things to a map) but uses JMS with MDBs.
I'd like to have all the systems use a pub/sub type design to get the messages (vice listening on a port).
My initial thoughth is to put a SOAP interface (using gSoap) in front of each of the legacy C++ apps. Define a webservice that invokes methods on the SOAP interface each time a message is passed to it (the webservice) from a MDB (via a JMS queue).
Does this make conceptual sense? I'm trying not to alter the legacy apps at all if possible.
As always, I appreciate everyone's feedback and necessary critcisim
20 years ago
Just registered with Promteric. Website wouldn't process the request but the helpful operator did.
I received an email with the voucher dated May 21st. Registration doesn't open until June 4th.
You got it. That's really the whole point of this; to allow QA to execute JUnit tests from their Test Management sofwtare they use for Functional and Performance tests.
20 years ago
Thanks for the replies. I guess in my project's case, we are trying to use the same common repsoitory that we use for functional and load tests (Rational TestManager).
20 years ago
Ok let's start simple. Let's say all I want to do is capture Pass/Fail results from my JUnit execution that is run on our nightly builds. Maybe this is like a stability metric for planning purposes.
Both the test tool(TestManager) and JUnit have a base class which is normally extended in your unit test creation. The TestManager base class sets up the connection to the repository for storing results. I have a simple method call in this base class to log a result.
What's the smartest way to do this integration? Have the JUnit base class extend the testManager base class and modify the Assert method in the JUnit framework to log a pass/fail result as well as display a JUnit failure?
20 years ago
Hey Guys-
I'm starting a little proof of concept effort and wanted to bounce some ideas off everyone.
I remeber not too long ago when unit testing "was not cool" in the developer ranks. The advent of JUnit has changed much of that. I have started to incorporate it's usage more and more into my team's development processes.
JUnit's lacks a few things however wehn we start talking about big projects. The first is the concept of a test repository where I can associate say a subsystem with it's related unit tests. Or a way to track the history of pass fail results for a certain component.
A lot of 3rd party tools provide this and I'd like to try some integration between the JUnit framework and Rational's TestManager as a starting point (since I have a license for it ).
I'd like to post a few questions as seperate threads as I go through this and get everyone's thoughts on approach. If nothing else, it puts a little more activity on this board
20 years ago
I think that the component icon (the square with the 2 rectangles) is supposed to symbolize the interface (that's the rectangles) with the implementation. Refer to the three amigos book on the UML to verify.
However if you *really* wanted the lollipop on your model you could define it in a class diagram somewhere and drag it onto your Component diagram.
HTH
[ May 08, 2003: Message edited by: John Pritchard ]
Having taken the 486 exam just a few days ago, I can tell you there are several questions worded like this on the exam although you won't find the word REQUIRED but as the you guys have stated, BEST.
BEST for me translated into "common and accepted practices". The "multiple use case sceanrios in a single interaction digram" answer for example; you coulddo it this way, but it is arguably not a common practice.
Just be very comfortable with the uses and purposes of the UML diagrams. You'll be given lots of scenario based quesions where they ask you which diagrams are best used.
Lastly be aware that some of the questions are subjective at best. At least that's how I am defending my 83% score
loading eclipse plug-ins now.
Really gives you a whole new appreciation for what can be done with notepad
20 years ago
Made some coffee
Read the headlines
perused the forum topics
still waiting
20 years ago
I was one of the lucky few who got a free test voucher last year from IBM. It was set to expire tomorrow so I had procrastinated long enough.
This test is not difficult if you have worked on any medium sized project that used Use Cases and did some modeling. I did not study for this exam but in retrospect, Larman's book is probably all that is needed.
Here's what I saw:
UML
- You only need to be comfortable with the symbology for aggregation/composition in class diagrams.
- You should know what an activity diagram looks like
- You should be able to read a sequence diagam with guard condictions and iterations
- Be very comfortable with what a deployment diagram is used for
Architecure
- I had a number of question on concurrency and which diagrams best support analysis/changes to concurrency requirements
- Several questions on MVC and which layer of objects is responsibel for what
- Be comfortable with looking at a use case, a corresponing class diagram, and knowing how/where to add new responsibilities to new and existing classes
Misc
- A few questions on how to organize your team around Use Cases and package hierarchies
- No questions on lifecycle phases
Overall not a difficult test. As another writer observed there are a few questions that appear subjective (at least that's how I'm defining my 83%