feat(WP-0016/C2-C3): close build infrastructure workplan

C2: document actual GHC cache behavior (-fbyte-code, no .o files, .hi
caching via -fwrite-interface) and correct CLAUDE.md cache claim.
C3: create IHUB-WP-0017 error-fix loop workplan (ralph-compatible,
5 tasks E1-E5, runs inside devenv shell).
WP-0016 status → done.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-04-10 23:36:06 +00:00
parent 469ed6c758
commit 34be62de04
3 changed files with 169 additions and 14 deletions

View File

@@ -79,11 +79,7 @@ scripts/compile-check --bg # same, background-friendly (no colour/title)
scripts/compile-check-core # Layer 1+2 only — verify clean base before touching Layer 3
```
**GHC object cache**: compiled `.o`/`.hi` files persist in `.devenv/` between restarts. After the first full build (2060 min on constrained hardware), subsequent single-module rebuilds take 530 seconds. Cache location:
```bash
find /home/tegwick/inter-hub -name "*.hi" 2>/dev/null | head -5
```
**GHC compilation mode**: IHP uses `-fbyte-code` — GHCi compiles modules to bytecode in memory, not to persistent `.o` files. The `-fwrite-interface` flag writes `.hi` type-info files alongside source modules; these survive session restarts and skip re-type-checking of unchanged modules. Each ghcid restart regenerates bytecode for the full module graph. Expect 2060 min on first build; restarts are somewhat faster due to `.hi` caching but not dramatically so. `.hi` files are written next to their source files (e.g. `Web/Controller/Sessions.hi`).
**Error-fix discipline**: fix bottom-up (Layer 1 → 4). Fix one module at a time; wait for ghcid to reload before moving on. See `workplans/IHUB-WP-0016-build-infrastructure-and-error-loop.md` for the full SOP.

View File

@@ -4,7 +4,7 @@ type: workplan
title: "Build Infrastructure: Incremental Compilation and Autonomous Error-Fix Loop"
domain: inter_hub
repo: inter-hub
status: open
status: done
owner: custodian
topic_slug: inter_hub
created: "2026-04-10"
@@ -71,6 +71,12 @@ skips the database startup, port binding, and service orchestration overhead.
### C1 — Create `scripts/compile-check` script ✓
```task
id: C1
status: done
priority: high
```
Also created `scripts/compile-check-core` and `.ghci-core` to compile Layer 1+2
in isolation (no Main.hs, no Layer 3). Resolves the architecture gap where Layer
2 correctness could not be verified independently.
@@ -103,19 +109,40 @@ This relies on `.ghci` (already present) which sets up `-j1` and loads the IHP c
### C2 — Verify GHC object cache persistence
GHC stores compiled objects in IHP's `.devenv` or a project-local `dist-newstyle/`
equivalent. Verify the cache survives between `devenv up` restarts:
```bash
# After first successful build, check cache location:
find /home/tegwick/inter-hub -name "*.o" -newer Main.hs 2>/dev/null | head -5
```task
id: C2
status: done
priority: medium
```
If the cache is cleared on each `devenv up` restart, add `IHP_BUILD_CACHE` or
configure the Nix build dir to persist. Document the location here.
**Finding (2026-04-10):** IHP's `applicationGhciConfig` sets `-fbyte-code`. GHCi compiles
modules to bytecode in memory — there are **no persistent `.o` object files**. Each ghcid
session recompiles all bytecode from source.
Our `-fwrite-interface` flag (C4) writes `.hi` interface files alongside source modules.
These survive session restarts and allow GHCi to skip re-type-checking unchanged modules.
This saves some time on restarts but does NOT eliminate full bytecode recompilation.
**Net result:** No persistent `.o` cache. `.hi` files provide partial speedup (skip
type-checking). A single-module change still triggers full bytecode regeneration of that
module and its dependents. For true persistent incremental builds, `-fobject-code` would
be needed — not worth adding given IHP's design tradeoffs.
**Action:** No configuration change needed. Expectation documented here and in CLAUDE.md.
### C3 — Autonomous error-fix loop (ralph-workplan)
```task
id: C3
status: done
priority: high
```
Delivered as `workplans/IHUB-WP-0017-error-fix-loop.md`. Run inside `devenv shell`:
```bash
/ralph-workplan workplans/IHUB-WP-0017-error-fix-loop.md
```
Configure a ralph-workplan loop:
1. `scripts/compile-check --bg` runs ghcid in background, writing to `/tmp/ihub-compile-errors.txt`
2. Monitor tool watches the log file for new error output
@@ -130,6 +157,12 @@ Configure a ralph-workplan loop:
### C4 — Add GHC `-fwrite-interface` and `-keep-going` flags to `.ghci` ✓
```task
id: C4
status: done
priority: high
```
```
-- Continue past errors in other modules (report all errors in one pass)
:set -fkeep-going
@@ -142,6 +175,12 @@ stopping at the first failure — critical for batch error fixing.
### C5 — Document module surface constraints in CLAUDE.md ✓
```task
id: C5
status: done
priority: medium
```
Added "Compilation Layers" section to CLAUDE.md covering:
- Never modify Layer 1 during error-fix loops (and why — cascade invalidation)
- Layer breakdown with file counts

View File

@@ -0,0 +1,120 @@
---
id: IHUB-WP-0017
type: workplan
title: "Autonomous Error-Fix Loop: Reach Clean Build"
domain: inter_hub
repo: inter-hub
status: active
owner: custodian
created: "2026-04-10"
updated: "2026-04-10"
depends_on: [IHUB-WP-0016]
---
# IHUB-WP-0017 — Autonomous Error-Fix Loop: Reach Clean Build
## Goal
Drive the inter-hub codebase to a clean `ghcid` build ("All good, N modules loaded")
by fixing compilation errors bottom-up, one module at a time.
## Pre-conditions
- Must be run inside `devenv shell` (IHP_LIB must be set, ghcid in PATH)
- WP-0016 complete (compile-check scripts and ghci flags in place)
## How to start
```bash
# Enter devenv shell first:
devenv shell
# Then start this loop:
/ralph-workplan workplans/IHUB-WP-0017-error-fix-loop.md
```
## Error-Fix SOP (Standard Operating Procedure)
```
1. Run: scripts/compile-check --bg
2. Tail /tmp/ihub-compile-errors.txt
3. Parse errors: group by module, order by Layer (1 → 4)
4. Fix Layer 1 errors first (rare; usually type/import issues in Web/Types.hs)
5. For each failing module in Layer 23:
a. Read current module state
b. Apply targeted fix (one module at a time)
c. Wait for ghcid "Reloading..." + result
d. If still failing: re-read error, apply next fix
e. If 3 attempts fail: escalate — flag_for_human with module name and error
6. After Layer 3 clean: fix Layer 4 (FrontController/Routes)
7. "All good (N modules)" in log = build clean
```
**Layer discipline** (never violate):
- Never modify Layer 1 (`Web/Types.hs`, `build/Generated/*.hs`) during error-fix loops
- Layer 1 changes invalidate ALL 59 controllers and 120 views simultaneously
- Fix within the layer where the error is reported; escalate if the fix requires Layer 1
## Tasks
### E1 — Start compile-check and capture initial error set
```task
id: E1
status: todo
priority: high
```
Run `scripts/compile-check --bg` and read `/tmp/ihub-compile-errors.txt`. List all
failing modules and their layers.
### E2 — Fix Layer 2 errors (Application/Helper/*.hs)
```task
id: E2
status: todo
priority: high
```
Fix all errors in `Application/Helper/*.hs`. Verify via ghcid reload. Mark done when
Layer 2 shows no errors.
### E3 — Fix Layer 3 errors (Web/Controller/*.hs and Web/View/**/*.hs)
```task
id: E3
status: todo
priority: high
```
Fix errors in controllers and views. Work module-by-module, bottom-up. Mark done when
Layer 3 shows no errors.
### E4 — Fix Layer 4 errors (Web/FrontController.hs, Web/Routes.hs)
```task
id: E4
status: todo
priority: medium
```
Fix wiring errors last. Mark done when `ghcid` reports "All good, N modules loaded".
### E5 — Commit clean build and close WP-0014/A1
```task
id: E5
status: todo
priority: medium
```
After "All good" state:
1. Commit all fixes with `fix(WP-0017): reach clean build`
2. Update `workplans/IHUB-WP-0014-pre-flight-deployment-gaps.md` A1 status to done
3. Mark this workplan done
## Exit Criteria
- `ghcid` reports "All good, N modules loaded" with no errors
- All fixes committed
- WP-0014/A1 closed