New OS, old compiler: never a great mix

OpenVMS fans are debating the merits of putting a legendary legacy compiler onto the newest version of the operating system. VMS Software Inc. has been developing an X86-64 version of OpenVMS for the last five years. 2020 marked the first year when significant releases surfaced at beta test sites. The underlying code for VMS has been a challenge to migrate. VSI’s work has injected hope into the VMS community, though. So much hope that the requests are beginning to surface for a migration of the DIBOL language to the newest version of VMS.

It’s easy to see why the most ardent users of VMS would want their favorite tools to run on this 9.2 version of the OS. Every legacy tool that moves across means savings over migration projects. DIBOL and DBL work on such a fundamental level that a port to the 9.2 release could mean many applications will survive.

The point of the VSI port, however, has been to give the legacy OS a newer infrastructure. For pure compatibility with newer hardware, virtualized Alpha and VAX solutions preserve everything. Tickets for the 9.2 train will be in high demand among application tool providers.

DIBOL going x86?

A lively discussion on the OpenVMS boards examines the prospects for seeing DIBOL running anywhere but an 8.x release of the OS. Stephen Hoffman summed up the possible outcomes for what’s a classic problem for new operating systems. Although OpenVMS is carrying a 9.2 version number when it becomes a reality, the world will be seeing an OS rewritten from the ground up.

“Absent a sufficiently large number of zeros included with the request, VSI won’t be writing the compiler for you,” Hoffman says. “They’ve done far too much work already.”

What makes the We Need DIBOL quest more quioxtic? The source code sits with a third party that’s been making DIBOL available to OpenVMS users for decades. Synergex is watching for interest from users who’d pay for a port to the 9.2 release. The likely future for a new DIBOL emerges in Hoffman’s summary of the challenge.

“If Synergex isn’t porting DIBOL/DBL, that can be inferred to believe the market here is small. That usually then means a custom compiler will be expensive [for VSI to port]. That’s for a product at risk of being undercut by some future port by Synergex. That’ll be an expensive compiler, for the work and for the risk involved.”

The stream of news about the new OpenVMS was steady during 2020. Hoffman reminds the OS fans that it remains early days in the life of native OpenVMS on Intel hardware. “It’s also some time before this OpenVMS x86-64 discussion is even particularly relevant. The second native-tooling beta V9.1 hasn’t arrived yet (early 2021 was predicted). The V9.2 production release will probably be a year or three after that.”

Moving things like DIBOL onto newer software architectures is costly. It’s just the sort of project a system vendor with resources like HP Enterprise might take on — if it were 2001, rather than 2021.

Leave a Reply