navicore/loglings
1
0
Fork 0
Code Issues Pull requests Releases Activity Actions

● Migration complete and verified. You were right to want to "find out" — it was a bit more than a rename. Here's #2

Merged
navicore merged 1 commit from migrate-plgc into main 2026-06-04 22:41:24 +00:00
Conversation 0 Commits 1 Files changed 8 +115 -88
navicore commented 2026-06-04 22:40:10 +00:00
Owner
Copy link

what we found and did.

What actually changed (3 things, all handled)

  1. Flag rename — --goal → --query. (src/runner.rs, tests/curriculum.rs)
  2. New runtime dependency: clang ≥ 15 — plgc is a real LLVM compiler; it compiles each exercise to a native
    binary and links it with clang. Already provisioned: the navicore-rust runner image has clang-19, your machine
    has clang-18.
  3. CI install — the crate is plg-compiler (binary plgc), and patch-prolog2 isn't on crates.io. Per your call:
    git-install, rev-pinned (cargo install --git …/patch-prolog2.git plg-compiler --rev 2e5f8ba --locked). Captured
    in the workflow + ROADMAP that crates.io is deliberately deferred until the system matures.

What stayed identical (so the curriculum needed zero content changes)

  • Exit-code contract 0/1/2/3 — byte-for-byte the same, so runner.rs's mapping is unchanged.
  • run subcommand + --format text.
  • The ISO-subset language at parity — every operator we teach works (=.., functor/arg, prefix/infix, right-assoc
    ^), :- dynamic works, and crucially still no op/3/postfix — so the 03-operators boundary lesson is still
    accurate (now an explicit exclusion in plgc's docs).
  • Latency is a non-issue: 0.12s per compile+run; the 54-compile semantic suite runs in 3.6s.

Verified

just ci is green — 6 curriculum tests, all 27 exercises compiled and passing under plgc, fmt/clippy clean,
release build ok. Files touched: runner.rs, tests/curriculum.rs, justfile, the workflow, README, ARCHITECTURE,
ROADMAP, and the design doc — all references now point at plgc/patch-prolog2 (with intentional history notes
where the old engine is mentioned).

The one thing I can't confirm from here is the git-install succeeding on the runner (network + repo read access)
— that'll show on the first CI run. The plgc --version line right after the install will make a failure
obvious. If the runner can't read patch-prolog2 over https, the fallback is baking plgc into the image (option 1
from earlier) — say the word and I'll switch it.

what we found and did. What actually changed (3 things, all handled) 1. Flag rename — --goal → --query. (src/runner.rs, tests/curriculum.rs) 2. New runtime dependency: clang ≥ 15 — plgc is a real LLVM compiler; it compiles each exercise to a native binary and links it with clang. Already provisioned: the navicore-rust runner image has clang-19, your machine has clang-18. 3. CI install — the crate is plg-compiler (binary plgc), and patch-prolog2 isn't on crates.io. Per your call: git-install, rev-pinned (cargo install --git …/patch-prolog2.git plg-compiler --rev 2e5f8ba --locked). Captured in the workflow + ROADMAP that crates.io is deliberately deferred until the system matures. What stayed identical (so the curriculum needed zero content changes) - Exit-code contract 0/1/2/3 — byte-for-byte the same, so runner.rs's mapping is unchanged. - run subcommand + --format text. - The ISO-subset language at parity — every operator we teach works (=.., functor/arg, prefix/infix, right-assoc ^), :- dynamic works, and crucially still no op/3/postfix — so the 03-operators boundary lesson is still accurate (now an explicit exclusion in plgc's docs). - Latency is a non-issue: 0.12s per compile+run; the 54-compile semantic suite runs in 3.6s. Verified just ci is green — 6 curriculum tests, all 27 exercises compiled and passing under plgc, fmt/clippy clean, release build ok. Files touched: runner.rs, tests/curriculum.rs, justfile, the workflow, README, ARCHITECTURE, ROADMAP, and the design doc — all references now point at plgc/patch-prolog2 (with intentional history notes where the old engine is mentioned). The one thing I can't confirm from here is the git-install succeeding on the runner (network + repo read access) — that'll show on the first CI run. The plgc --version line right after the install will make a failure obvious. If the runner can't read patch-prolog2 over https, the fallback is baking plgc into the image (option 1 from earlier) — say the word and I'll switch it.
● Migration complete and verified. You were right to want to "find out" — it was a bit more than a rename. Here's
All checks were successful
CI - Linux / CI - Linux x86_64 (pull_request) Successful in 39s
Details
e8552f8c96 
what we found and did.

  What actually changed (3 things, all handled)

  1. Flag rename — --goal → --query. (src/runner.rs, tests/curriculum.rs)
  2. New runtime dependency: clang ≥ 15 — plgc is a real LLVM compiler; it compiles each exercise to a native
  binary and links it with clang. Already provisioned: the navicore-rust runner image has clang-19, your machine
  has clang-18.
  3. CI install — the crate is plg-compiler (binary plgc), and patch-prolog2 isn't on crates.io. Per your call:
  git-install, rev-pinned (cargo install --git …/patch-prolog2.git plg-compiler --rev 2e5f8ba --locked). Captured
  in the workflow + ROADMAP that crates.io is deliberately deferred until the system matures.

  What stayed identical (so the curriculum needed zero content changes)

  - Exit-code contract 0/1/2/3 — byte-for-byte the same, so runner.rs's mapping is unchanged.
  - run subcommand + --format text.
  - The ISO-subset language at parity — every operator we teach works (=.., functor/arg, prefix/infix, right-assoc
  ^), :- dynamic works, and crucially still no op/3/postfix — so the 03-operators boundary lesson is still
  accurate (now an explicit exclusion in plgc's docs).
  - Latency is a non-issue: 0.12s per compile+run; the 54-compile semantic suite runs in 3.6s.

  Verified

  just ci is green — 6 curriculum tests, all 27 exercises compiled and passing under plgc, fmt/clippy clean,
  release build ok. Files touched: runner.rs, tests/curriculum.rs, justfile, the workflow, README, ARCHITECTURE,
  ROADMAP, and the design doc — all references now point at plgc/patch-prolog2 (with intentional history notes
  where the old engine is mentioned).

  The one thing I can't confirm from here is the git-install succeeding on the runner (network + repo read access)
  — that'll show on the first CI run. The plgc --version line right after the install will make a failure
  obvious. If the runner can't read patch-prolog2 over https, the fallback is baking plgc into the image (option 1
  from earlier) — say the word and I'll switch it.
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
navicore/loglings!2
No description provided.