# lib/vm/src — Sleela OOD equivalents of the VM C/C++ *Idea System(TM) / Idea(TM). From Author; distributive.* This directory holds the **Sleela object-oriented equivalents** of the native C/C++ VM code, so a user can **select a VM other than the Native VM in `/impl`**. On system start — or during a conference with the user — the native VM loads these `.sleela` into memory, lets the user pick a version, and then loads that version's C/C++ `main`. ## The two sides | Native C/C++ | Sleela OOD equivalent (here) | Mirrors | |---|---|---| | `vm_config.{h,c}` (v001) | `SLVmConfig.sleela` | memory/thread/verbosity/capability tuning | | `vm_report.{h,c}` (v001) | `SLVmReport.sleela` | uniform `[idea-vm] SEV (sub/code): msg` | | `vm_status.{h,c}` (v001) | `SLVmStatus.sleela` | the "what am I running" banner | | `vm_memory.{h,c}` (v002) | `SLVmMemory.sleela` | pressure watermarks | | `vm_result.{h,c}` (v003) | `SLVmResult.sleela` | typed results + legacy-status map | | `vm_telemetry.{h,c}` (v004) | `SLVmTelemetry.sleela` | counters + table saturation | | `vm_gc.{h,c}` (v005) | `SLVmGc.sleela` | GC pressure/pacing quality | | `vm_profile.{h,c}` (v006) | `SLVmProfile.sleela` | dispatch profiler | | `vm_capability.{h,c}` (v007) | `SLVmCapability.sleela` | capability enforcement | | `vm_platform.{h,c}` (v008) | `SLVmPlatform.sleela` | platform parity | | `vm_audit.{h,c}` (v009) | `SLVmAudit.sleela` | hash-chained audit trail | | `vm_runtime.{h,c}` (v010) | `SLVmRuntime.sleela` | the consolidated facade | | `lib/vm/src/sleela_vm_source.c` | `SLVmSource.sleela` | VM source-requirement validation | Plus the selection layer: - `SLVmVersion.sleela` — base manifest a selectable version extends. - `virtual-machine/vm-version/NNN/SLVmVersionNNN.sleela` — the per-version manifest (id, label, native dir + `main`, composed units, config profile). - `virtual-machine/vm-version/SLVmRegistry.sleela` — the catalogue; `list()`, `select(n)`, `boot(n)`. ## Selecting a VM (start / conference / terminal) ``` SLVmRegistry reg = new SLVmRegistry(); reg.load(); // load all ten manifests into memory reg.list(); // enumerate for the user SLVmRuntime rt = reg.boot(10); // choose v010 (consolidated); builds its runtime ``` Conceptually: `sleela vm list` / `sleela vm select ` from the terminal, or a setup/conference choice. The chosen manifest's `build()` returns a configured `SLVmRuntime`; its `nativeMain` names the C/C++ entry the native loader then builds and boots. ## Opcode seam VM selection routes through the **`OP_VM_SELECT` (148)** service opcode — the Sleela-side `SLVMNativeBinding` VM-exit the host services by loading the chosen version's native main. The source ISA (`lib/vm/InstructionSet.sleela`), the native enum (`impl/core/sleela_core.h` + its copies), the native dispatch (`impl/core/sleela_core.c`), and `OPCODE-MAP.md` were all updated together per the opcode maintenance rule — this change also closed the prior `OP_IC_*` (126–147) source-ISA lag. ## Design notes - Authored once here (shared classes), with a thin self-describing manifest in each of the ten `vm-version/NNN/` folders — the OOD-respectful design, not ten copies of identical class bodies. - Match the lib/vm Sleela conventions: `#sleela 1.3`, the five-field doc-header, `configure()`-style init (no constructors), `int` state machines, and numbered fields instead of arrays (the surface has no arrays). - Every file is grammar-clean through the real `sleela` compiler. Two method names were adjusted because `record` is contextual after `.` in call position (`SLVmAudit.log`/`logResult`, `SLVmProfile.attribute`).