It's often crucial to know the manner in which users are interacting with a distributed component or application. With a little time and effort it's possible to create a universal logging object that can be used in any distributed PowerBuilder component or application that will automate the task of generating such information. This article discusses the creation of such an object.

When you need 'today's date' in a client/server application, or a Web application for that matter, what date do you use? I had this discussion recently with Ed Foshko at Fidelity Investments. You might think the answer is easy - just use the appropriate function to get the date. Life as usual isn't that simple. You have at least three choices: the local computer, an application server or a database. Let's say you have an application in which you wish to capture a user's ID along with the date and time a record was updated. When the update occurs, make an entry in a logging table of some sort. This approach should enable you to determine who modified a record last...but does it?

Last year, after migrating a client's application from PFC 5.0.02 to 6.5, then applying some enhancements to take advantage of new features, the program began giving the following PFC error message: 'For certain functionality the PFC DataStore requires a reference to its parent window. One of these cases has been encountered. To let the DataStore know who its parent window is, call the of_SetParentWindow() function after the DataStore creation.'

Okay, so you've used showplan, statistics i/o and statistics time to identify how your query has been optimized and what it's costing you. That's a good starting point. Sometimes all the answers are right there in the query plan or resource counts. They'll point out a table scan, the fact that your query is using a different index than you expected or that most of your I/O relates to one particular table or SQL statement. You know you have an index on the search columns, so why did it choose a tablescan? You've done your homework and identified that one index is more selective than another, yet the optimizer decided to access the table with the less selective index. You suspect that one join order is better than another, based on your knowledge of the data and relationships, but it didn't choose that join order. Unfortunately, these tools won't help you in these situations because they fail to answer the question 'Why?' They only answer the questions 'How?' and 'How much?'

The application server market continues to be one of the hottest areas in technology. Why? Because an application server is the central piece of software in a Web-based architecture. An application server centralizes application logic and data access, making it possible to access applications over the Internet or an intranet (see my article in the May '99 PBDJ [Vol. 6, issue 5]).

Tt's a new year, a new millennium (give or take a year), and we've just released version 3.5 of the Sybase Enterprise Application Server. There's no better time to contemplate where we've been and where we're going.

It's often crucial to know the manner in which users are interacting with a distributed component or application. With a little time and effort it's possible to create a universal logging object that can be used in any distributed PowerBuilder component or application that will automate the task of generating such information. This article discusses the creation of such an object.