Hi everyone,
I experimented how far I could get migrating an Oracle Forms application to Vaadin and Spring boot with Claude Code. I used Oracle’s Summit demo and focused on the ORDERS form.
After building a default architecture, I tried to create a generic workflow: Extract technical rules from the Forms sources → Combine the technical rules and architecture to “specifications” → split the specifications to tickets → Implement the tickets → Validate the result (and improve the loop)
This is still a small experiment. Summit doesn’t exercise stateful PL/SQL packages or dynamic SQL, and the next iteration’s stricter traceability checks haven’t been implemented yet.
Full write-up · Code and migration rules
What do you think about this approach? Have you done Oracle to Java migration and got bitten by some features that are tough to migrate?
I did a Forms to Vaadin migration. But this was not technical transformation but kind of a reimplementation of many Forms concepts with Vaadin.
With AI simple technology transformations are not sustainable. I do all modernizations projects using SDD.
Old app → Specs (use cases, entity model etc) → new app
For the experiment I used Matt Pocock skills and create a specification based on the technical findings and what I wanted to keep or how I wanted to migrate the view. This ticket was created : PRD: Functional Vaadin ORDERS view (S_ORD/S_ITEM master-detail spine) [v3] · Issue #77 · jcgueriaud1/forms-summit-migration · GitHub then detailed in multiple implementations tickets.
It needs time to create the proper spec. ( And I spent more time on trying to extract the needed information from the forms to challenge the analyst that shapes the specifications).
This spec, once implemented, is not kept since the code become the source of truth.
I could keep it but I don’t like the fact that there is nothing that forces a synchronization between spec and code.
I also think that when the specifications are becoming as precise as the code then maybe the code is only needed.
I definitely need to buy and read your book to get a better idea of your methodology. And apply it for a experiment or a real project.
1 Like