2.0 KiB
Legacy compatibility and pilot contract
Required user-facing behavior
| Legacy behavior | Preserve in Next | Pilot status |
|---|---|---|
| Product/project selection workflow | preserve after separate UX analysis | not implemented |
| Plant 3D version discovery | preserve through adapter/override | foundation adapter |
| Local corporate plugin availability | preserve through valid local cache | designed/tested locally |
| P_Rec commands and Ribbon | preserve through one managed entry assembly | plugin-owned, not launched here |
| Immediate legacy return | preserve LauncherMode=Legacy |
designed |
| Other legacy plugins | retain existing chain until individually migrated | required |
Can be removed only for migrated P_Rec_Annot
- OrthoLoader configuration fan-out and twenty sequential
NETLOADcalls. - Generated LISP/BAT as the package-loading mechanism, once explicit Next startup is runtime-proven.
- Per-file legacy copy replacing an active module directory.
They remain necessary for other legacy corporate plugins until those plugins have their own migration and acceptance evidence.
Duplicated or avoidable work
The source shows installer recursive copy/delete, updater hash/copy work, generated LISP generation and twenty sequential module loads. The precise overlap and cost require measurement. Next reduces avoidable work by comparing manifests, reusing a valid local version, staging only a new version and loading one entry assembly. It must not claim that legacy runs these actions on every AutoCAD startup without runtime evidence.
P_Rec_Annot pilot
The pilot is a dry-run contract only:
discover Plant 3D 2026 -> validate fixture manifest -> local test-source staging
-> activate version pointer -> build one startup NETLOAD script
-> construct ProcessStartInfo -> do not start process
It uses a fixture manifest with a synthetic SHA-256 and a test-created placeholder file. No real bundle, Plant assembly, package cache or production catalog is included.