It's not just mainframes that are at risk. PC-based client-server applications are not compliant -- up to 20% of logic of these systems.
Data imported from spreadsheets or other PC's are suspect. If these data are corrupt, they will infect the compliant mainframes (if any).
Banks are especially vulnerable.
This is from GOVERNMENT COMPUTER NEWS (March 9).
* * * * * * *
Just 20 months away from 2000, it's dawning on systems managers that client-server networks might pose even more insidious date code problems than the mainframe systems they have concentrated on fixing. . . .
Client-server networks account for at least 50 percent of the software portfolios in many organizations, and fixing those networks and applications will take much more effort than anyone realizes, said Mike Smith, product manager for year 2000 tool vendor McCabe & Associates Inc. of Columbia, Md.
For one thing, high-level languages such as C, C++, Ada, Visual Basic and PowerBuilder used in client-server applications are inherently more complex than Cobol used for mainframe apps, Smith said.
"We're finding about 5 percent of the lines of code in client-server applications will be affected, but almost 20 percent of the logic in those applications will be affected," he said.
Another reason is the lack of standards for client-server programming, which could lead to widespread corruption of data that passes among apps. . . .
When data travels among desktop applications, the problems are every bit as complicated as on a mainframe, Arora said, especially when incompatible century windows are present. . . .
"If you move data from one application to another, you're in danger of coming up with different answers" for the same calculation, Arora said.
Many users download mainframe data containing two-digit dates and manipulate it in spreadsheets or databases before sending it out in reports "with very little checking and cross-checking," [William] Ulrich said.
The data integrity problem is by no means confined to the government's decision-support systems. An average investment bank, for example, does more than $300 billion in daily transactions, relying on PCs for correct calculations and processing.