Research index
These reports are the evidence base for the advanced architecture. They are research input,
not ADRs; the decisions are in ../spec/architecture.md §2.2.
| Report | What it answers |
|---|---|
indy-hardware.md | The Indy’s CPU, caches, TLB, HPC3, SCSI, Seeq, HAL2, Newport, Vino, DS1386, GIO and PROM. |
rust-x11-research.md | The Rust X11 ecosystem, what a minimal IRIX-capable server needs, and the build-vs-port verdict. |
native-window-system-research.md | Every realistic native window-system option, with the bandwidth arithmetic. |
rust-fs-research.md | Rust filesystems, the XFS v4 on-disk format, I7FS and the storage architecture. |
plan9-rust-research.md | Plan 9’s ideas, Rust 9P, 9P over WAN, capabilities, MCP and content-addressed images. |
erofs-big-endian-research.md | Whether to fork EROFS to big-endian (verdict: do not). |
compressed-ram.md | What Meta’s CRAM actually is and whether compressed RAM fits the Indy (verdict: no). |
rust-os-design-research.md | A broad survey of Rust OS design choices. |
rust-std-irix7.md | What std for IRIX 7 means, the native-vs-compat split, the POSIX.1-2024 native libc ABI it requires, and the implementation and PR plan. |
rust-std-build-spike.md | The measured spike: build the std rlib for the native n32 target from a patched rust-src and libc, the exact method and patch set, the result, and the remaining runtime gaps. |
std-hello-bringup.md | The first linked and running static std image: the 42-symbol shim, kernel syscall additions, loader fixes, output and remaining POSIX gaps. |
relibc-n32-bringup.md | The relibc n32/QEMU gate: patched relibc, TLS, setjmp, pthreads, and the provisional (linked, not booted) IRIX PAL scaffold. |
quic-primary.md | Whether QUIC should replace TCP as the native transport, the Indy’s crypto/code/memory budget, the recommended QUIC profile and the measured spike that decides it. |
The reports cite primary sources; they are kept verbatim (including their own spelling) for traceability.