Freedom of target
Java, C#, C++ or a domain platform. The front end is independent of the destination, so the choice is yours.
Somewhere in your estate sits a system written in a language with no vendor, no community and no tool market: an in-house 4GL, a scripting layer someone brilliant built in 1994, a controller language that runs your plants. Every conversion tool on the market fails on line 1, because none of them can parse it.
This is the problem Modernize Software was built for. DMS, our symbolic AI and enhanced compiler technology, is configurable to any language. We define yours from your manuals and your code, then translate it automatically to a modern platform.
Trusted By
Dow's factories worldwide ran on MOD5 controllers programmed in Dowtran, a language Dow created and only Dow used. Rewriting by hand across 1,000 systems risked the process knowledge that keeps chemical plants safe.
Dow contracted Semantic Designs in 2012 to build Sequoia, a custom DMS-based tool that recovers the process control models buried in Dowtran code and carries them to modern industrial controllers, preserving the hard-won chemical process rules. Dow has extended the contract repeatedly as the migration rolls across its plants.
Grammar and semantics are captured from compiler manuals, language guides and your real code corpus.
A DMS front end for your language is built and validated until it reads your entire codebase cleanly.
Translation rules map your language's constructs to the target platform, tuned on your actual programs.
Legacy and translated systems run side by side on your cases until equivalence is demonstrated.
SabreTalk, the airline industry's PL/I dialect, has its own dedicated page: SabreTalk migration.
Commercial 4GLs whose vendors have faded fit the same machinery: if there's a grammar to recover and code to learn from, a translator can be built.
Java, C#, C++ or a domain platform. The front end is independent of the destination, so the choice is yours.
Business and process rules embedded in the old language are extracted and carried across, the asset your rewrite would have gambled.
Your teams keep maintaining the legacy system until final conversion runs on a fresh snapshot.
The translator can be kept and rerun, valuable when migrations roll out site by site over years, as Dow's does.
General questions
No. Documentation speeds up grammar capture, but the code corpus itself is the strongest evidence of what the language accepts. We've reconstructed language definitions from code, fragments of manuals and interviews with the remaining experts.
DMS was designed for exactly this: it's a configurable analysis and translation engine, and each language is a configuration. Defining a new front end is standard practice for us, done for dozens of languages over 3 decades.
In-house 4GLs, report and scripting languages built inside one company, controller languages like Dowtran, and commercial languages whose vendors are gone. If nobody sells a compiler for it anymore, it qualifies.
A rewrite needs people who understand both the old language and the buried business rules, the exact people you're running out of. Automated translation extracts and preserves those rules instead of re-deriving them from memory.
A custom tool that reverse-engineers Dowtran control programs into explicit process models, which Dow forward-engineers onto modern industrial controllers. Over 300 plant systems had moved by 2020, with the contract extended repeatedly since 2012.
Yes. Real proprietary codebases contain generations of style and extension. The front end is validated against your full corpus, so drift shows up early and gets built into the grammar.
Through behaviour, on your systems. The legacy and translated versions run side by side on your test cases and production scenarios, and every difference traces to a specific translation rule that gets fixed once, everywhere.
You end up owning modern, maintainable code and, where the engagement is structured that way, continued use of the migration tool for future waves. Ownership and licensing are settled in the agreement.
A language that exists in one company is the worst possible case for guesswork. DMS using its symbolic AI technology translates deterministically from a defined grammar, and Gen AI assists around the edges, on documentation and comprehension.
A free assessment: send us sample code and whatever documentation exists. We evaluate grammar recoverability, size the front-end build and give you a phased plan before any commitment.
Send sample code for a free assessment of your proprietary or 4GL system.
.png)
info@semanticdesigns.com
