From 1660c16f4f296dee8c1efc972d4671208ed05a8d Mon Sep 17 00:00:00 2001 From: zinc Date: Wed, 14 Oct 2026 15:41:56 +0330 Subject: xref: checked-in baseline, --check self-test hooked into the lsp section, todo --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'README.md') diff --git a/README.md b/README.md index 0644d4e..35e94b6 100644 --- a/README.md +++ b/README.md @@ -38,7 +38,7 @@ make js make run PROG=examples/hello.zc ``` -the cores embed their genome with c23 `#embed`, so the c compiler has to be clang 19 or gcc 15 or later. `src/tools/findcc.sh` finds one (`CC` set on the command line or in the environment is used as given, and must support it too). `make genome` bootstraps the compiler, compares two self-compiled images, and writes `build/core/genome.bin` only after they agree; `make` reruns it whenever a source of the genome changes, and the cores that embed it are rebuilt after it. `make check` runs the smaller suite while iterating. `make parity` (`src/tools/parity.sh`, `--help` for options) compiles the test and example programs with zc0 and with the self-hosted compiler and compares what they print, which the fixpoint check cannot see; programs zc0 rejects are skipped and counted, intended differences are listed in `tests/parity.known`, and work files go to `build/parity/`. `make xref` (`src/tools/xref.sh`, `--help` for options) lists, for every top-level variable of the compiler, the functions that write it and the functions that only read it, in `build/xref/report.txt` and `report.json`; the exact scoping comes from the language server's `textDocument/references` on the joined compiler source, so a local that shadows a global is not counted, and a write is an `=`, a compound assignment, `++` or `--`, an indexed store, or `&x`. `src/tools/xref.sh --check` (or `make xref XREF_ARGS=--check`) exits 1 when a global gains a writer that `tests/xref.baseline` does not list, so a change that lets a new function write core compiler state shows up in review; `--update` rewrites the baseline. the language server is interpreted without the arm64 jit, so a run takes minutes. `ONBOARD.md` is a reading and verification order for getting to know the code. +the cores embed their genome with c23 `#embed`, so the c compiler has to be clang 19 or gcc 15 or later. `src/tools/findcc.sh` finds one (`CC` set on the command line or in the environment is used as given, and must support it too). `make genome` bootstraps the compiler, compares two self-compiled images, and writes `build/core/genome.bin` only after they agree; `make` reruns it whenever a source of the genome changes, and the cores that embed it are rebuilt after it. `make check` runs the smaller suite while iterating. `make parity` (`src/tools/parity.sh`, `--help` for options) compiles the test and example programs with zc0 and with the self-hosted compiler and compares what they print, which the fixpoint check cannot see; programs zc0 rejects are skipped and counted, intended differences are listed in `tests/parity.known`, and work files go to `build/parity/`. `make xref` (`src/tools/xref.sh`, `--help` for options) lists, for every top-level variable of the compiler, the functions that write it and the functions that only read it, in `build/xref/report.txt` and `report.json`; the exact scoping comes from the language server's `textDocument/references` on the joined compiler source, so a local that shadows a global is not counted, and a write is an `=`, a compound assignment, `++` or `--`, an indexed store, or `&x`. `src/tools/xref.sh --check` (or `make xref XREF_ARGS=--check`) exits 1 when a global gains a writer that `tests/xref.baseline` does not list, so a change that lets a new function write core compiler state shows up in review; `--update` rewrites the baseline. the language server is interpreted without the arm64 jit, so a run takes tens of seconds (about 10 s with 8 servers on the box). `ONBOARD.md` is a reading and verification order for getting to know the code. ### command-line help -- cgit v1.2.3