feat: split js_image_layer into fine grained layers

[gregmagolan]

Resolves #1641.

This change splits the output of js_image_layer into 5 layers:

1. node layer contains Node.js and the node fs patches
2. package_store_3p layers containers all 3p deps in the .aspect_rules_js package store
3. package_store_1p layers containers all 1p deps in the .aspect_rules_js package store
4. node_modules layer contains all node_modules/foo symlinks into the packages tore
5. app layer contains everything that doesn't fall into any of the above layers

[aspect-workflows]

Test

All tests were cache hits

190 tests (100.0%) were fully cached saving 51s.

---

Test

e2e/bzlmod

All tests were cache hits

4 tests (100.0%) were fully cached saving 467ms.

---

Test

e2e/gyp_no_install_script

All tests were cache hits

2 tests (100.0%) were fully cached saving 693ms.

---

Test

e2e/js_image_oci

All tests were cache hits

1 test (100.0%) was fully cached saving 9s.

---

Test

e2e/npm_link_package

All tests were cache hits

2 tests (100.0%) were fully cached saving 845ms.

---

Test

e2e/npm_link_package-esm

All tests were cache hits

2 tests (100.0%) were fully cached saving 773ms.

---

Test

e2e/npm_translate_lock

All tests were cache hits

1 test (100.0%) was fully cached saving 73ms.

---

Test

e2e/npm_translate_lock_empty

All tests were cache hits

1 test (100.0%) was fully cached saving 73ms.

---

Test

e2e/npm_translate_lock_multi

All tests were cache hits

2 tests (100.0%) were fully cached saving 271ms.

---

Test

e2e/npm_translate_lock_partial_clone

All tests were cache hits

1 test (100.0%) was fully cached saving 117ms.

---

Test

e2e/npm_translate_lock_subdir_patch

All tests were cache hits

1 test (100.0%) was fully cached saving 248ms.

---

Test

e2e/npm_translate_package_lock

All tests were cache hits

1 test (100.0%) was fully cached saving 73ms.

---

Test

e2e/npm_translate_yarn_lock

All tests were cache hits

1 test (100.0%) was fully cached saving 73ms.

---

Test

e2e/package_json_module

All tests were cache hits

1 test (100.0%) was fully cached saving 511ms.

---

Test

e2e/pnpm_lockfiles

All tests were cache hits

6 tests (100.0%) were fully cached saving 2s.

---

Test

e2e/pnpm_workspace

All tests were cache hits

8 tests (100.0%) were fully cached saving 3s.

---

Test

e2e/pnpm_workspace_rerooted

All tests were cache hits

6 tests (100.0%) were fully cached saving 2s.

---

Test

e2e/repo_mapping

All tests were cache hits

2 tests (100.0%) were fully cached saving 451ms.

---

Test

e2e/rules_foo

All tests were cache hits

2 tests (100.0%) were fully cached saving 393ms.

---

Test

e2e/vendored_node

All tests were cache hits

1 test (100.0%) was fully cached saving 187ms.

---

Buildifier      Format

[thesayyn]

I think we should put node and node_modules into same layer as those are less likely to change.

[thesayyn]

hmm how about; node_modules_links + package_store_3p + node they all go into the same layer, as they won't change frequently?

[gregmagolan]

hmm how about; node_modules_links + package_store_3p + node they all go into the same layer, as they won't change frequently?

They could; node_modules_links include 1p links but won't change as much as the actual contents of the 1p links. Is there a downside to more fine grained layers?

[thesayyn]

Other than network round trip not really, more the better, but i am afraid that we are splitting it too much that it will be confusing.

[gregmagolan]

Other than network round trip not really, more the better, but i am afraid that we are splitting it too much that it will be confusing.

Its transparent for users of rules_oci tho right since tars = [":layers",], in oci_image. Just the rules_docker example has them listed explicitly and we don't care about rules_docker ease-of-use since that rule set is abandonware now.

[thesayyn]

Makes sense, i guess this is fine, fine grained layering will give users more power.

[matthewjh]

Completely agree with fine-grained layering along these lines - it should save a lot of bandwidth and time for orgs that have 100+ services.

Thanks @gregmagolan!

[thesayyn]

I think this PR has a bad rebase, @gregmagolan could you fix that before i can take a look?

[gregmagolan]

I think this PR has a bad rebase, @gregmagolan could you fix that before i can take a look?

Ahh yes, the base branch was rebased which messed up the commits. All clean again now.