Files
el/lang/AGENTS.md
T

145 lines
5.6 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# El Language — Agent Guide
El is a self-hosting, statically-typed language that compiles to C. This file orients agents that work on El itself or on programs written in El.
---
## What El Is
El compiles `.el` source → C → native binary. Every El value is `el_val_t` (int64_t). Strings are heap pointers cast through int64_t. The compiler is written in El (self-hosting).
**The compiler pipeline:**
```
elc-cli.el
└─ imports: compiler.el
└─ imports: lexer.el, parser.el, codegen.el, codegen-js.el
```
The canonical compiler binary is `dist/platform/elc`. It was produced by running an earlier version of itself on `elc-cli.el`.
---
## The Two Layers — Know Which One You're In
### Layer 1: El programs (`.el` files)
This is where almost all work belongs. El programs are source files that get compiled by `elc`. New library functions, application logic, and language-level utilities all go here as `.el` files.
**Do not add C code when El can express it.** If functionality can be built from existing El primitives (string ops, `exec`, `fs_read/write`, `http_post`, etc.), write it in El.
### Layer 2: The C seed (`el-compiler/runtime/el_seed.c`)
This is the self-contained C OS-boundary layer. It provides the `__`-prefixed primitives that compiled El programs call: libcurl HTTP, pthreads, filesystem I/O, arena allocation, etc. It is **not generated** — it is maintained by hand.
The old `el_runtime.c` has been archived to `el-compiler/runtime/legacy/`. The runtime is now native El (`runtime/*.el`). `el_seed.c` replaces `el_runtime.c` as the sole C compilation dependency.
**Only edit `el_seed.c` when you genuinely need OS-level access** (raw sockets, GPU calls, new libcurl features). For everything else, write El.
When you do add a C builtin:
1. Add the C function to `el_seed.c`
2. Declare it in `el_seed.h`
3. Add it to the `builtin_arity` table in `el-compiler/src/codegen.el` (so the compiler knows the arg count)
4. Rebuild the elc binary (see below)
---
## Rebuilding the Compiler
After changing any `.el` source in `el-compiler/src/`:
```bash
cd /Users/will/Development/neuron-technologies/foundation/el
./dist/platform/elc elc-cli.el > elc-new.c
cc -std=c11 -I el-compiler/runtime -lcurl -lpthread \
-o dist/platform/elc-new \
elc-new.c el-compiler/runtime/el_seed.c
# Verify self-hosting:
./dist/platform/elc-new elc-cli.el > elc-verify.c
diff elc-new.c elc-verify.c # should be identical
mv dist/platform/elc-new dist/platform/elc
```
After changing `el_seed.c` only (no El source changes), rebuild downstream programs but do NOT need to rebuild the compiler binary itself — the seed is linked at the application level, not the compiler level.
---
## How El Programs Are Built
Each El application has a `build.sh` that:
1. Concatenates all `.el` source files (stripping `import` lines)
2. Runs `elc` to produce a `.c` file
3. Runs `cc` linking against `el_seed.c`
Example (cgi-studio daemon):
```bash
cd products/cgi-studio/el-daemon
./build.sh
```
When you add a new `.el` file to an application, add it to that application's `build.sh` concat list.
---
## Parallelism in El
El is single-threaded at the application level. Parallelism is achieved through subprocess fan-out:
```el
// Pattern: write payloads to temp files, exec bash script with & and wait,
// read results back from temp files.
fn http_post_parallel(urls: [String], bodies: [String]) -> [String] {
// ... bash fan-out via exec() ...
}
```
Use `exec()` (blocking) or `exec_bg()` (fire-and-forget) with shell scripts to run concurrent work. There is no goroutine or async/await — parallelism goes through the OS process layer.
---
## Key Files
| Path | What it is |
|------|-----------|
| `dist/platform/elc` | Canonical compiler binary (arm64 Mac) |
| `el-compiler/src/codegen.el` | Code generator — builtin arity table lives here |
| `el-compiler/src/lexer.el` | Lexer |
| `el-compiler/src/parser.el` | Parser |
| `el-compiler/runtime/el_seed.c` | Self-contained C OS-boundary layer (replaces el_runtime.c) |
| `el-compiler/runtime/el_seed.h` | Seed header (C function declarations) |
| `spec/language.md` | Language specification |
| `BOOTSTRAP.md` | How to recover the compiler from scratch |
| `elc-cli.el` | Compiler entry point |
| `elc-combined.el` | Pre-merged single-file compiler (used during early bootstrap) |
---
## HTTP Timeout
The El HTTP client (libcurl) defaults to **60 seconds**. Override per-process via `EL_HTTP_TIMEOUT_MS` env var. Set it before spawning any subprocess that makes long API calls:
```el
exec("EL_HTTP_TIMEOUT_MS=300000 " + SOME_BIN + " " + args + " 2>&1")
```
---
## Operator naming convention (for cognitive `.el` modules)
When you write `.el` that names a **cognitive faculty / operator** (the language
faculty, appreciation, interoception, the summon loop under `elp/`), name it for
its **functional human equivalent** — the faculty a mind would name — **not** its
linear-algebra operation. Put the math characterization in the `@impl`
doc-comment, never in the operator's public name (e.g. public `discern`/`contrast`
`@impl subtract (ab)`; public `summon`/`recall``@impl LOCAL nearest-region
+ bounded spreading activation`). Full table + rationale in
`foundation/el/AGENTS.md` § *Operator naming convention*. This does **not** apply
to plain library/compiler code (a `sort` is a `sort`).
## Rules
- New library functions → write in El
- New OS/hardware primitives → write in C and register in `codegen.el` arity table
- Never edit `dist/platform/elc` directly — always rebuild from source
- Never modify `el_seed.c` to add functionality that El can express
- Cognitive-faculty `.el` → name for the mind, algebra in `@impl` (see above)