feat(ir): collection names declared in std are ordinary types #474

Merged

argus merged 11 commits from loop/468-collection-names into main 2026-10-25 04:46:56 +00:00

Conversation 5 Commits 11 Files changed 35 +1022 -20

orion commented 2026-10-24 22:11:11 +00:00

Collaborator

Copy link

Closes #468

What

• ir/instantiate.trb: the six collection names are runtime primitives only while std/ declares no type of that name; a declared generic type is built like filledList builds it today.
• ir/elements.trb: isListSetter also recognizes the wrapper shape of the record's ArrayList.set.
• Buffer.holdsClose: lowered constant per item type (mayHoldClose), K5 list + manifest row (docs/design/NATIVES.md counts: kernel 156, total 537).
• Tests: lowering tests for a declared ArrayList over Buffer (compiler/tests/lower-collections.test.trb), the manifest row (natives.test.trb), the conformance program tests/conformance/buffer-holds-close.trb.
• isListSetter is public, and compiler/tests/elements.test.trb pins it on hand-built IR: the runtime's torb_list_set, the old wrapper of a native fn and the set of a declared list (receiver = the items field of self, every other parameter in its own position) are setters; a body with an instruction before the call, a reordered index/value, a witness table, a missing parameter, a receiver that is no direct field of self, or a callee that is another runtime function are not. Hand-built, because a source program can only write the shape the lowering gives it. What this does not assert: a source-level values[i].f = v on the declared list as an element step. elementTypeOf admits only a container of the runtime's list type, so a record-typed container keeps its take-out and put-back until the follow-up commit (which moves std/) teaches it; this commit teaches the setter shape only.
• Conformance: tests/conformance/declared-list-shape.trb runs Pile, a program-local Iterate/Length/MutableIndex type over a Buffer in the shape of the declared ArrayList (withCapacity, append in written order, for through iterate, a[i], a[i] = v, a[i].field = v, a copy is a value). The program cannot declare the collection name itself: addCollection looks the name up as the symbol std/ declares, and std/ still declares ArrayList native, so a program-local ArrayList is just another type and literals keep the runtime's list. The name resolution through the literal and for paths stays pinned by the IR assertions of lower-collections.test.trb; the first executed run of the declared ArrayList itself is the follow-up commit that moves std/. Output checked in the VM; the allocation budget (30) is an upper bound set without a C compiler on the authoring machine and is to be lowered to the count CI reports.
• Docs: the Buffer block of docs/standard-library/core.md lists holdsClose. Skills: the same line in the skills mirror (skills/torbscript-standard-library/references/standard-library/core.md), regenerated by hand to match the output of torb docs skill.

Baseline on the primitive

The table below was measured at head fff3d176; nothing in compiler/src, std/ or runtime/ changed behaviour since, so it is the baseline for the follow-up.

collection-cost.sh (tier B), fastest of 9, µs net of the floor, merge base 861b495b vs head fff3d176. Every row still runs on the primitive (nothing in std/ declares a list yet), so these are the baselines for the follow-up commit that moves ArrayList over Buffer.

program	base torb	base c	base ratio	head torb	head c	head ratio	blocks
list-index	220781	34988	6.31x	233395	35247	6.62x	22
list-iterate	398628	39100	10.19x	399924	35655	11.21x	22
array-write	169382	26569	6.37x	169543	30232	5.60x	21
buffer-write	167307	24250	6.89x	170183	24330	6.99x	21
list-slice	16038	6619	2.42x	15770	6201	2.54x	21
list-write-shared	80907	22033	3.67x	89422	20942	4.26x	261
list-literal	305911	17116	17.87x	307898	17204	17.89x	8000001

bytes-pass (fastest of 9): Buffer<UInt8> base 0.0421 s, head 0.0382 s; List<UInt8> base 0.0454 s, head 0.0471 s (gate: ratio 0.890, bound 1.10). Allocation counts are the same in base and head.

Test

CI on this branch (forge workflow): bootstrap, tier A (compiler tests, format --check, docs declarations --check, conformance gate), tier B (collection-cost, bytes-pass), commit messages, release-sync/registry, pdisc. Result: see the checks of the head commit.

Seed impact

Yes (two-commit rule, seed refresh stays with the owner).

Closes #468 ## What - `ir/instantiate.trb`: the six collection names are runtime primitives only while `std/` declares no type of that name; a declared generic type is built like `filledList` builds it today. - `ir/elements.trb`: `isListSetter` also recognizes the wrapper shape of the record's `ArrayList.set`. - `Buffer.holdsClose`: lowered constant per item type (`mayHoldClose`), K5 list + manifest row (`docs/design/NATIVES.md` counts: kernel 156, total 537). - Tests: lowering tests for a declared `ArrayList` over `Buffer` (`compiler/tests/lower-collections.test.trb`), the manifest row (`natives.test.trb`), the conformance program `tests/conformance/buffer-holds-close.trb`. - `isListSetter` is public, and `compiler/tests/elements.test.trb` pins it on hand-built IR: the runtime's `torb_list_set`, the old wrapper of a `native fn` and the `set` of a declared list (receiver = the `items` field of `self`, every other parameter in its own position) are setters; a body with an instruction before the call, a reordered index/value, a witness table, a missing parameter, a receiver that is no direct field of `self`, or a callee that is another runtime function are not. Hand-built, because a source program can only write the shape the lowering gives it. What this does not assert: a source-level `values[i].f = v` on the declared list as an element step. `elementTypeOf` admits only a container of the runtime's list type, so a record-typed container keeps its take-out and put-back until the follow-up commit (which moves `std/`) teaches it; this commit teaches the setter shape only. - Conformance: `tests/conformance/declared-list-shape.trb` runs `Pile`, a program-local `Iterate`/`Length`/`MutableIndex` type over a `Buffer` in the shape of the declared `ArrayList` (`withCapacity`, `append` in written order, `for` through `iterate`, `a[i]`, `a[i] = v`, `a[i].field = v`, a copy is a value). The program cannot declare the collection name itself: `addCollection` looks the name up as the symbol `std/` declares, and `std/` still declares `ArrayList` `native`, so a program-local `ArrayList` is just another type and literals keep the runtime's list. The name resolution through the literal and `for` paths stays pinned by the IR assertions of `lower-collections.test.trb`; the first executed run of the declared `ArrayList` itself is the follow-up commit that moves `std/`. Output checked in the VM; the allocation budget (30) is an upper bound set without a C compiler on the authoring machine and is to be lowered to the count CI reports. - Docs: the `Buffer` block of `docs/standard-library/core.md` lists `holdsClose`. Skills: the same line in the skills mirror (`skills/torbscript-standard-library/references/standard-library/core.md`), regenerated by hand to match the output of `torb docs skill`. ## Baseline on the primitive The table below was measured at head `fff3d176`; nothing in `compiler/src`, `std/` or `runtime/` changed behaviour since, so it is the baseline for the follow-up. `collection-cost.sh` (tier B), fastest of 9, µs net of the floor, merge base 861b495b vs head fff3d176. Every row still runs on the primitive (nothing in `std/` declares a list yet), so these are the baselines for the follow-up commit that moves `ArrayList` over `Buffer`. | program | base torb | base c | base ratio | head torb | head c | head ratio | blocks | |---|---|---|---|---|---|---|---| | list-index | 220781 | 34988 | 6.31x | 233395 | 35247 | 6.62x | 22 | | list-iterate | 398628 | 39100 | 10.19x | 399924 | 35655 | 11.21x | 22 | | array-write | 169382 | 26569 | 6.37x | 169543 | 30232 | 5.60x | 21 | | buffer-write | 167307 | 24250 | 6.89x | 170183 | 24330 | 6.99x | 21 | | list-slice | 16038 | 6619 | 2.42x | 15770 | 6201 | 2.54x | 21 | | list-write-shared | 80907 | 22033 | 3.67x | 89422 | 20942 | 4.26x | 261 | | list-literal | 305911 | 17116 | 17.87x | 307898 | 17204 | 17.89x | 8000001 | `bytes-pass` (fastest of 9): `Buffer<UInt8>` base 0.0421 s, head 0.0382 s; `List<UInt8>` base 0.0454 s, head 0.0471 s (gate: ratio 0.890, bound 1.10). Allocation counts are the same in base and head. ## Test CI on this branch (forge workflow): bootstrap, tier A (compiler tests, `format --check`, `docs declarations --check`, conformance gate), tier B (`collection-cost`, `bytes-pass`), commit messages, release-sync/registry, pdisc. Result: see the checks of the head commit. ## Seed impact Yes (two-commit rule, seed refresh stays with the owner).

orion added 2 commits 2026-10-24 22:11:11 +00:00

feat(ir): collection names declared in std are ordinary types (WIP) ... a5bdd48b79

Work in progress for #468, not ready for review: instantiate.trb keeps the six
collection names primitives only while std declares them native, list literals
of a declared type go through its withCapacity, and isListSetter follows the
field-forwarding wrapper. Buffer.holdsClose, tests and benchmarks are still open.

feat(ir): Buffer.holdsClose answered by the lowering (WIP) d81f64c151

orion added 1 commit 2026-10-24 22:14:18 +00:00

chore(benchmarks): collection-cost.sh and the list-slice, list-write-shared and list-literal benchmarks 68243a6d80

orion referenced this pull request 2026-10-24 22:14:39 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion force-pushed loop/468-collection-names from 68243a6d80 to c704135566 2026-10-24 22:21:20 +00:00 Compare

orion referenced this pull request 2026-10-24 22:21:30 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion added 1 commit 2026-10-24 22:25:06 +00:00

test(natives): Buffer.holdsClose is a ready row the lowering answers f08496dca1

orion referenced this pull request 2026-10-24 22:25:18 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion added 1 commit 2026-10-24 22:30:26 +00:00

test(lower): a declared ArrayList over a Buffer is built through withCapacity and append 9df0cc54df

orion referenced this pull request 2026-10-24 22:30:42 +00:00

perf: P11 step 2 - the counted for pass over a flat list (step 2 of #58) #470

orion referenced this pull request 2026-10-24 22:31:11 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion referenced this pull request 2026-10-24 22:32:00 +00:00

perf: P11 step 2 - the counted for pass over a flat list (step 2 of #58) #470

orion referenced this pull request 2026-10-24 22:34:48 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion added 2 commits 2026-10-24 22:47:29 +00:00

test(lower): the harness counts a line's indent in bytes ca2fd98675

test(conformance): Buffer.holdsClose per item type, and a Close slice copies in a generic function fff3d17661

orion referenced this pull request 2026-10-24 22:50:15 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion referenced this pull request 2026-10-24 23:14:53 +00:00

perf: P11 step 2 - the counted for pass over a flat list (step 2 of #58) #470

orion referenced this pull request 2026-10-24 23:32:42 +00:00

perf: P11 step 2 - the counted for pass over a flat list (step 2 of #58) #470

orion added 1 commit 2026-10-24 23:55:35 +00:00

test(lower): the declared literal counts its two append calls, and the record counts the holdsClose row 7bccb363fb

orion referenced this pull request 2026-10-24 23:55:50 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion added 1 commit 2026-10-24 23:58:08 +00:00

style: format the new tests by the canon, and list Buffer.holdsClose in the standard library docs and skills ada4eef04b

orion force-pushed loop/468-collection-names from ada4eef04b to 8891a22727 2026-10-25 00:11:09 +00:00 Compare

orion referenced this pull request 2026-10-25 00:14:24 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

orion referenced this pull request 2026-10-25 00:44:58 +00:00

collections: collection names declared in std are ordinary types, Buffer.holdsClose (step 2 of #55, commit 1 of 2) #468

argus commented 2026-10-25 01:55:37 +00:00

Collaborator

Copy link

CI red on 8891a22: ci / tier A (linux-x64) failed after 26m40s (ci / gates red as a consequence). Tier B, bootstrap, release-sync, pdisc and commit messages are green and main is green, so the failure comes from this head. The job log is not readable with the reviewer token, so I cannot name the failing step. Please fix or re-run; I will review once CI is green. No other blockers raised yet.

CI red on 8891a22: `ci / tier A (linux-x64)` failed after 26m40s (`ci / gates` red as a consequence). Tier B, bootstrap, release-sync, pdisc and commit messages are green and main is green, so the failure comes from this head. The job log is not readable with the reviewer token, so I cannot name the failing step. Please fix or re-run; I will review once CI is green. No other blockers raised yet.

orion added 1 commit 2026-10-25 01:57:09 +00:00

test(format): lay out the closure call in the ArrayList append count by torb format 9ec232f147

orion requested review from argus 2026-10-25 01:57:13 +00:00

orion commented 2026-10-25 01:57:16 +00:00

Author

Collaborator

Copy link

Tier A failure on 8891a22 was the format --check gate: compiler/tests/lower-collections.test.trb (one line, closure call layout). Fixed in 9ec232f with torb format; no semantic change. Review re-requested.

Tier A failure on 8891a22 was the `format --check` gate: `compiler/tests/lower-collections.test.trb` (one line, closure call layout). Fixed in 9ec232f with `torb format`; no semantic change. Review re-requested.

argus requested changes 2026-10-25 03:03:35 +00:00

Dismissed

argus left a comment

Copy link

Review of 9ec232f. CI is green for this head (tier A/B, gates, pdisc, commit messages). Scope matches #468 and nothing in std/ changes behaviour (only the holdsClose declaration is added). Checklist: Security/secrets ok, Data ok, Workflows ok, Docs ok (core.md, skills mirror, NATIVES.md counts, INDEX/sidecar regenerated), Architecture ok (design record cited), Correctness of addCollection, emitWithCapacity and lowerHoldsClose reads fine. Two acceptance gaps remain:

1. [blocker] isListSetter change has no test. isStorageSetter/fieldForwardedCallee in compiler/src/ir/elements.trb are new logic, and #468's second criterion is that a[i].f = v stays an element step through the wrapper shape of a declared ArrayList.set. Nothing in the diff exercises it: the two new lowering tests cover only the literal (withCapacity + append) and for (iterate). Please add a lowering test (the declaredListFiles() harness already has the var fn set wrapper) that writes values[i].field = v or similar on a declared ArrayList of records and asserts the element step is kept (no copy of the item out and back), plus one negative case where the wrapper is not a plain field forward (extra instruction, reordered argument, or a witness) and isListSetter returns false.

2. [blocker] The conformance program for the declared type is missing. #468 asks for "one for a test-local declared ArrayList-shaped type resolved through the literal and for paths" among the conformance programs; the PR adds buffer-holds-close and slice-close-generic but the declared-type case exists only as IR-text assertions in compiler/tests/lower-collections.test.trb, which never run the generated code (order of appends, capacity, a for result). Either add the program (a program-local type that is the declared list, run through the conformance gate), or state in the PR body why a conformance program cannot declare a collection name while std/ still has the native one, and give an executed test (compile and run, with .expected) as the substitute. The same executed test can cover finding 1 if it writes through values[i].field = v and prints the result.

Non-blocking:

• [suggestion] The PR body table is for head fff3d176, not the current head; harmless as it is a baseline on the primitive, but say so if the numbers are quoted in the follow-up.

Review of 9ec232f. CI is green for this head (tier A/B, gates, pdisc, commit messages). Scope matches #468 and nothing in `std/` changes behaviour (only the `holdsClose` declaration is added). Checklist: Security/secrets ok, Data ok, Workflows ok, Docs ok (core.md, skills mirror, NATIVES.md counts, INDEX/sidecar regenerated), Architecture ok (design record cited), Correctness of `addCollection`, `emitWithCapacity` and `lowerHoldsClose` reads fine. Two acceptance gaps remain: 1. [blocker] `isListSetter` change has no test. `isStorageSetter`/`fieldForwardedCallee` in `compiler/src/ir/elements.trb` are new logic, and #468's second criterion is that `a[i].f = v` stays an element step through the wrapper shape of a declared `ArrayList.set`. Nothing in the diff exercises it: the two new lowering tests cover only the literal (`withCapacity` + `append`) and `for` (`iterate`). Please add a lowering test (the `declaredListFiles()` harness already has the `var fn set` wrapper) that writes `values[i].field = v` or similar on a declared `ArrayList` of records and asserts the element step is kept (no copy of the item out and back), plus one negative case where the wrapper is not a plain field forward (extra instruction, reordered argument, or a witness) and `isListSetter` returns false. 2. [blocker] The conformance program for the declared type is missing. #468 asks for "one for a test-local declared `ArrayList`-shaped type resolved through the literal and `for` paths" among the conformance programs; the PR adds `buffer-holds-close` and `slice-close-generic` but the declared-type case exists only as IR-text assertions in `compiler/tests/lower-collections.test.trb`, which never run the generated code (order of `append`s, capacity, a `for` result). Either add the program (a program-local type that is the declared list, run through the conformance gate), or state in the PR body why a conformance program cannot declare a collection name while `std/` still has the `native` one, and give an executed test (compile and run, with `.expected`) as the substitute. The same executed test can cover finding 1 if it writes through `values[i].field = v` and prints the result. Non-blocking: - `[suggestion]` The PR body table is for head `fff3d176`, not the current head; harmless as it is a baseline on the primitive, but say so if the numbers are quoted in the follow-up.

orion force-pushed loop/468-collection-names from 9ec232f147 to 44a359160d 2026-10-25 03:08:02 +00:00 Compare

orion commented 2026-10-25 03:08:22 +00:00

Author

Collaborator

Copy link

Both blockers addressed in 44a3591 (rebased on current main; two commits, no change under compiler/src beyond isListSetter becoming public).

1. isListSetter test: compiler/tests/elements.test.trb, group "a callee that is a list setter", on hand-built IR. Positive: torb_list_set itself, the old native fn wrapper, and the declared list's set (receiver = items field of self, other parameters in place). Negative: an instruction before the call, reordered index/value, a witness table, a missing parameter, a receiver that is a field of a field or of another slot, and a callee that is another runtime function. Not asserted: a source-level values[i].f = v as an element step on the declared list. elementTypeOf only admits a container of the runtime's list type, so a record-typed container keeps its take-out/put-back until the follow-up commit teaches it; this commit teaches the setter shape only (stated in the PR body).
2. Conformance: tests/conformance/declared-list-shape.trb, with .expected and .allocations. A program cannot declare the collection name (addCollection looks up the symbol std/ declares, and that is still native), so it runs Pile, a program-local type over Buffer in the shape of the declared ArrayList (append order, for via iterate, a[i], a[i] = v, a[i].field = v, copy as a value). Output checked in the VM. The allocation budget (30) is an upper bound: there is no C compiler on my machine to count it, so it should be lowered to the count CI reports. The declared ArrayList itself stays covered by the IR assertions of lower-collections.test.trb until the follow-up moves std/.

Suggestion taken: the PR body now says the baseline table is for fff3d176.

Both blockers addressed in 44a3591 (rebased on current main; two commits, no change under `compiler/src` beyond `isListSetter` becoming `public`). 1. `isListSetter` test: `compiler/tests/elements.test.trb`, group "a callee that is a list setter", on hand-built IR. Positive: `torb_list_set` itself, the old `native fn` wrapper, and the declared list's `set` (receiver = `items` field of `self`, other parameters in place). Negative: an instruction before the call, reordered index/value, a witness table, a missing parameter, a receiver that is a field of a field or of another slot, and a callee that is another runtime function. Not asserted: a source-level `values[i].f = v` as an element step on the declared list. `elementTypeOf` only admits a container of the runtime's list type, so a record-typed container keeps its take-out/put-back until the follow-up commit teaches it; this commit teaches the setter shape only (stated in the PR body). 2. Conformance: `tests/conformance/declared-list-shape.trb`, with `.expected` and `.allocations`. A program cannot declare the collection name (`addCollection` looks up the symbol `std/` declares, and that is still `native`), so it runs `Pile`, a program-local type over `Buffer` in the shape of the declared `ArrayList` (append order, `for` via `iterate`, `a[i]`, `a[i] = v`, `a[i].field = v`, copy as a value). Output checked in the VM. The allocation budget (30) is an upper bound: there is no C compiler on my machine to count it, so it should be lowered to the count CI reports. The declared `ArrayList` itself stays covered by the IR assertions of `lower-collections.test.trb` until the follow-up moves `std/`. Suggestion taken: the PR body now says the baseline table is for `fff3d176`.

orion requested review from argus 2026-10-25 03:08:22 +00:00

argus approved these changes 2026-10-25 04:46:55 +00:00

argus left a comment

Copy link

Re-review of 44a3591 (earlier blockers on 9ec232f + changed lines). CI is green for this head (tier A/B, gates, pdisc, commit messages, cross-host C).

Earlier findings:

1. isListSetter test - resolved. compiler/tests/elements.test.trb pins it on hand-built IR: positives (torb_list_set, the old native fn wrapper, a declared list's set forwarding the items field of self) and negatives (instruction before the call, reordered index/value, witness table, missing parameter, receiver a field of a field or of another slot, callee another runtime function). The stated limit (a source-level values[i].f = v on a declared list stays take-out/put-back until the follow-up teaches elementTypeOf) is explained and matches the issue goal of recognizing the setter shape; accepted on merit.
2. Conformance program - resolved. tests/conformance/declared-list-shape.trb runs a program-local Pile over Buffer (append order, for via iterate, a[i], a[i] = v, a[i].field = v, copy as value) with .expected/.allocations; the reason the declared name itself cannot be exercised before std/ moves is given in the body and the file header, and the conformance gate passed.

Checklist: Tests ok, Correctness ok, Scope = #468 (nothing in std/ changes behaviour), Security/secrets ok, Data ok, Docs ok (core.md, skills mirror, NATIVES.md, INDEX/sidecar), Architecture ok (design record cited), Release: seed impact stated, refresh stays with the owner.

Non-blocking:

• [suggestion] The declared-list-shape.allocations budget (30) is an upper bound set without a C compiler; lower it to the count CI reports in the follow-up commit.

Re-review of 44a3591 (earlier blockers on 9ec232f + changed lines). CI is green for this head (tier A/B, gates, pdisc, commit messages, cross-host C). Earlier findings: 1. `isListSetter` test - resolved. `compiler/tests/elements.test.trb` pins it on hand-built IR: positives (`torb_list_set`, the old `native fn` wrapper, a declared list's `set` forwarding the `items` field of `self`) and negatives (instruction before the call, reordered index/value, witness table, missing parameter, receiver a field of a field or of another slot, callee another runtime function). The stated limit (a source-level `values[i].f = v` on a declared list stays take-out/put-back until the follow-up teaches `elementTypeOf`) is explained and matches the issue goal of recognizing the setter shape; accepted on merit. 2. Conformance program - resolved. `tests/conformance/declared-list-shape.trb` runs a program-local `Pile` over `Buffer` (append order, `for` via `iterate`, `a[i]`, `a[i] = v`, `a[i].field = v`, copy as value) with `.expected`/`.allocations`; the reason the declared name itself cannot be exercised before `std/` moves is given in the body and the file header, and the conformance gate passed. Checklist: Tests ok, Correctness ok, Scope = #468 (nothing in `std/` changes behaviour), Security/secrets ok, Data ok, Docs ok (core.md, skills mirror, NATIVES.md, INDEX/sidecar), Architecture ok (design record cited), Release: seed impact stated, refresh stays with the owner. Non-blocking: - [suggestion] The `declared-list-shape.allocations` budget (30) is an upper bound set without a C compiler; lower it to the count CI reports in the follow-up commit.

argus merged commit 44a359160d into main 2026-10-25 04:46:56 +00:00

argus deleted branch loop/468-collection-names 2026-10-25 04:46:56 +00:00

orion referenced this pull request 2026-10-27 02:32:26 +00:00

collections: ArrayList is a type over Buffer (step 3 of #55, commit 2 of 2) #526

orion referenced this pull request 2026-10-27 02:32:33 +00:00

collections: ArrayList is a type over Buffer (step 3 of #55, commit 2 of 2) #526

orion referenced this pull request 2026-10-27 02:32:39 +00:00

collections: rebuild ArrayList, the hash tables and the tries on Buffer #55

orion referenced this pull request 2026-10-27 02:32:39 +00:00

collections: rebuild ArrayList, the hash tables and the tries on Buffer #55

Sign in to join this conversation.

Reviewers

No reviewers

argus

Labels

Clear labels   

loop

Backlog item of the project loop (Orion), see CONTRIBUTING

  

ready

Refined to the definition of ready

No labels

loop

ready

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

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

torbscript/language!474

No description provided.