Father of ASCII tackles Y2K

A new company, BigiSoft Inc., is marketing a tool that attacks Year 2000 problems at the level of ones and zeros, rather than in source code, and plans to bring this tool to the federal market.

Vertex 2000, created by Bob Bemer, who is known as the father of ASCII, is a departure from source code solutions for Cobol and other mainframe languages that use the techniques of windowing and horizontal expansion to make sure the computer will recognize the new millennium.

Vertex 2000 attacks the problem in the bowels of the system by working at the object code level. Object code is the intermediate stage between source code, written by a programmer, and low-level machine language, which actually carries out instructions.

Vertex 2000 scans object code for instructions that operate on years. It finds those instructions and makes changes automatically or by user prompts similar to a search-and-replace utility in a word processing program.

When the computer industry first started fixing Year 2000 problems, no one could find a way to do object code remediation, said Mike Dillard, vice president of BigiSoft. Consequently, most mainframe repair projects are relying on source code fixes using windowing or expansion, he said.

Windowing modifies the logic in the program so that the program decides what century a year belongs in based on its relationship to a "pivot" year. A pivot year of 68, for example, means two-digit numbers greater than 68 are treated as if they occur before 2000. Years less than 68 are treated as if they occur after 2000.

Expansion, which involves expanding date fields in source code and in data, requires modification of all code or else it will not run when the data is converted.

Dillard said that with the millennium now less than 500 days away, he would not try to convince managers of Year 2000 projects that are well under way with either technique to switch to Vertex 2000. But he said he would encourage them to use the product to test their work. Testing source code remediation is the most time-consuming part of a Year 2000 fix.

"[Agencies] can now have an independent assessment of what they've done," said Dillard, who expects most initial sales of Vertex 2000 to be for testing projects.

Bemer, 78, came out of retirement to create Vertex 2000, which he says will have a "profound consequence and application on corporate America and the world." Bemer, who in addition to leading the effort to establish ASCII as a universal character set for text files also played a role in the creation of Cobol, said Vertex 2000 can correct mainframe code 10 times faster than windowing and horizontal expansion.

"Any method of Year 2000 conversion that targets source code is little more than a vain attempt to speed up the hunt-and-peck approach to millennium compliance. It's still too time-consuming and costly," Bemer said.

Bemer founded the Richardson, Texas-based company BMR Software Inc. a little more than a year ago to develop and sell Vertex 2000. In August the company was bought and renamed BigiSoft after the patented name of Bemer's creation, the Bigit Methodology.

Bemer was named senior scientist and a member of the executive team of the Dallas-based BigiSoft, which also has offices in Atlanta and Orlando and which plans to open an office in Northern Virginia to support its efforts to promote Vertex 2000 to government customers, Dillard said.

Bemer's product uses a theory that is founded on good principles, said Leon A. Kappelman, a professor at the University of North Texas and the co-chairman of the Society for Information Management's Year 2000 working group. But how it works in different environments with different applications remains to be seen, Kappelman said.

"People who have Cobol running on a mainframe should evaluate its applicability in their environments," Kappelman said. It might save time, but if it means there will be severe performance implications, "then you either have to compensate by throwing more hardware at it, or you have to understand those considerations."

Analyst Stephanie Moore of Giga Information Group Inc., Westport, Conn., described the solution as a patch that makes the code impossible to maintain or recompile without destroying all the changes.

"It's an interesting solution for people who are desperate, kind of at the 11th hour, but in terms of it being a responsible way to fix your code and be able to maintain it going forward into the 21st century, it's not the best idea," Moore said. "It's more like a contingency or backup plan."

Last month BigiSoft executives made an initial foray into the government market with visits to several government resellers, including Science Applications International Corp.

"There are a number of good people in government trying to solve the problem, but the older products that do windowing and horizontal expansion are too slow," Dillard said. "If we continue along that path, the government will not be ready."