BE-267: Use bump scopes instead of full resets across transformation passes #8234

indietyp · 2026-01-01T17:59:01Z

🌟 What is the purpose of this PR?

Replaces the ResetAllocator pattern with scoped checkpoints in the HashQL MIR transformation pipeline. Instead of resetting the entire scratch allocator between passes, we now use checkpoint()/rollback() semantics and scoped sub-arenas, which provides more granular memory management, and enables more advanced patterns in the future.

🔍 What does this change?

Adds Checkpoint type and checkpoint()/rollback() methods to the BumpAllocator trait
Updates PreInlining pass to use BumpAllocator::scoped() for each sub-pass instead of resetting the allocator
Updates all transformation passes (CopyPropagation, CfgSimplify, InstSimplify, ForwardSubstitution, AdministrativeReduction, DeadStoreElimination, etc.) to work with scoped allocators
Updates benchmarks to use the new scoping pattern
Uses heap-based tracking for changed state in transformation passes

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

This PR:

does not modify any publishable blocks or libraries, or modifications do not need publishing

📜 Does this require a change to the docs?

The changes in this PR:

are internal and do not require a docs change

🕸️ Does this require a change to the Turbo Graph?

The changes in this PR:

do not affect the execution graph

🛡 What tests cover this?

Existing MIR transformation tests and compiletest UI tests

❓ How to test this?

Run cargo nextest run --package hashql-mir
Run transformation benchmarks: `cargo bench --package

cursor · 2026-01-01T17:59:06Z

PR Summary

Implements scoped bump allocation and updates MIR transforms to use it for finer-grained memory management.

Core allocators: Add Checkpoint and checkpoint()/rollback() to BumpAllocator; implement in Allocator, AllocatorScope, Heap, and Scratch; forward impls for &mut A.
MIR passes: Migrate AdministrativeReduction, CfgSimplify, CopyPropagation, InstSimplify, DeadStoreElimination, DeadLocalElimination, and SsaRepair to depend on BumpAllocator and use alloc.scoped(...) instead of reset(); remove reset calls and restructure helpers (e.g., mark_dead_blocks).
PreInlining driver: Run each sub-pass in a scope; track per-body Changed state using heap storage; global pass AdministrativeReduction also scoped.
ForwardSubstitution: Generalize to Allocator (default Global) instead of Scratch.
Benchmarks/compiletests: Update to pass Scratch through closures and reset per-iteration; adapt pass constructors to new_in(scratch).
Collections: WorkQueue::new_in now pre-allocates with VecDeque::with_capacity_in(domain_size, ...) to reduce reallocations.

^{Written by Cursor Bugbot for commit cb4fa90. This will update automatically on new commits. Configure here.}

indietyp · 2026-01-01T17:59:50Z

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

This stack of pull requests is managed by Graphite. Learn more about stacking.

augmentcode · 2026-01-01T18:05:30Z

🤖 Augment PR Summary

Summary: Refactors the HashQL MIR transformation pipeline to stop doing full scratch allocator resets between passes and instead rely on scoped bump allocation/checkpoints for finer-grained temporary memory management.

Changes:

Adds a Checkpoint type plus BumpAllocator::checkpoint()/rollback(), implemented for Allocator, AllocatorScope, Heap, and Scratch
Updates major MIR passes (e.g. CfgSimplify, CopyPropagation, DSE, SSARepair, AdministrativeReduction) to accept BumpAllocator and use alloc.scoped(...) instead of calling reset()
Refactors CfgSimplify internals into helper functions and runs sub-passes under allocator scopes
Adjusts PreInlining to scope each sub-pass and uses a heap-backed per-body changed-state slice for the fixpoint loop
Updates transformation benchmarks to thread a reusable Scratch through each iteration and construct passes with new_in(scratch)
Minor allocation improvement: WorkQueue now pre-allocates its VecDeque with domain-sized capacity

Technical Notes: The intent is to enable more granular allocator usage patterns (checkpoint/rollback + sub-arenas) while keeping pass-local allocations short-lived and composable.

_{🤖 Was this summary useful? React with 👍 or 👎}

augmentcode

Review completed. 1 suggestions posted.

Comment augment review to trigger a new review at any time.

libs/@local/hashql/mir/benches/transform.rs

codspeed-hq · 2026-01-01T18:06:08Z

CodSpeed Performance Report

Merging this PR will not alter performance

_{Comparing bm/be-267-hashql-do-not-reset-the-scratch-space-but-instead-use-scopes (cb4fa90) with bm/be-266-hashql-implement-pre-inlining-fix-point-loop (99146b5)¹}

Summary

✅ 17 untouched benchmarks
🗄️ 12 archived benchmarks run²

No successful run was found on bm/be-266-hashql-implement-pre-inlining-fix-point-loop (720ee51) during the generation of this report, so 54e9310 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩
12 benchmarks were run, but are now archived. If they were deleted in another branch, consider rebasing to remove them from the report. Instead if they were added back, click here to restore them. ↩

codecov · 2026-01-01T18:06:57Z

Codecov Report

❌ Patch coverage is 79.71530% with 57 lines in your changes missing coverage. Please review.
✅ Project coverage is 64.02%. Comparing base (720ee51) to head (cb4fa90).

Files with missing lines	Patch %	Lines
.../hashql/mir/src/pass/transform/cfg_simplify/mod.rs	87.83%	15 Missing and 8 partials ⚠️
libs/@local/hashql/core/src/heap/allocator.rs	0.00%	12 Missing ⚠️
libs/@local/hashql/core/src/heap/bump.rs	0.00%	8 Missing ⚠️
libs/@local/hashql/core/src/heap/mod.rs	0.00%	6 Missing ⚠️
libs/@local/hashql/core/src/heap/scratch.rs	0.00%	6 Missing ⚠️
...hql/mir/src/pass/transform/forward_substitution.rs	0.00%	2 Missing ⚠️

Additional details and impacted files

@@                                     Coverage Diff                                     @@
##           bm/be-266-hashql-implement-pre-inlining-fix-point-loop    #8234       +/-   ##
===========================================================================================
- Coverage                                                   81.20%   64.02%   -17.19%     
===========================================================================================
  Files                                                          82      729      +647     
  Lines                                                       10744    61511    +50767     
  Branches                                                      283     3643     +3360     
===========================================================================================
+ Hits                                                         8725    39383    +30658     
- Misses                                                       1912    21658    +19746     
- Partials                                                      107      470      +363

Flag	Coverage Δ
apps.hash-ai-worker-ts	`1.40% <ø> (?)`
apps.hash-api	`0.00% <ø> (?)`
local.hash-graph-sdk	`10.88% <ø> (?)`
local.hash-isomorphic-utils	`0.00% <ø> (?)`
rust.hash-graph-api	`2.89% <ø> (?)`
rust.hashql-ast	`87.25% <ø> (?)`
rust.hashql-compiletest	`46.65% <ø> (ø)`
rust.hashql-core	`81.63% <3.03%> (?)`
rust.hashql-eval	`68.54% <ø> (?)`
rust.hashql-hir	`89.10% <ø> (?)`
rust.hashql-mir	`87.63% <89.91%> (+0.04%)`	⬆️
rust.hashql-syntax-jexpr	`94.05% <ø> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

libs/@local/hashql/mir/benches/transform.rs

github-actions · 2026-01-02T16:57:49Z

Benchmark results

@rust/hash-graph-benches – Integrations

policy_resolution_large

Function	Value	Mean	Flame graphs
resolve_policies_for_actor	user: empty, selectivity: high, policies: 2002	$$26.3 \mathrm{ms} \pm 172 \mathrm{μs}\left({\color{gray}0.298 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: low, policies: 1	$$3.21 \mathrm{ms} \pm 13.0 \mathrm{μs}\left({\color{gray}0.101 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: medium, policies: 1001	$$11.7 \mathrm{ms} \pm 72.2 \mathrm{μs}\left({\color{gray}0.108 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: high, policies: 3314	$$41.4 \mathrm{ms} \pm 279 \mathrm{μs}\left({\color{gray}-0.275 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: low, policies: 1	$$13.7 \mathrm{ms} \pm 77.1 \mathrm{μs}\left({\color{gray}0.650 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: medium, policies: 1526	$$22.7 \mathrm{ms} \pm 132 \mathrm{μs}\left({\color{gray}-0.900 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: high, policies: 2078	$$29.7 \mathrm{ms} \pm 179 \mathrm{μs}\left({\color{lightgreen}-29.003 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: low, policies: 1	$$3.59 \mathrm{ms} \pm 13.4 \mathrm{μs}\left({\color{lightgreen}-82.009 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: medium, policies: 1033	$$13.4 \mathrm{ms} \pm 83.3 \mathrm{μs}\left({\color{lightgreen}-50.930 \mathrm{\%}}\right) $$	Flame Graph

policy_resolution_medium

Function	Value	Mean	Flame graphs
resolve_policies_for_actor	user: empty, selectivity: high, policies: 102	$$3.62 \mathrm{ms} \pm 15.0 \mathrm{μs}\left({\color{gray}-1.301 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: low, policies: 1	$$2.82 \mathrm{ms} \pm 13.1 \mathrm{μs}\left({\color{gray}-0.477 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: medium, policies: 51	$$3.17 \mathrm{ms} \pm 14.9 \mathrm{μs}\left({\color{gray}-1.210 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: high, policies: 269	$$4.95 \mathrm{ms} \pm 25.0 \mathrm{μs}\left({\color{gray}0.600 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: low, policies: 1	$$3.36 \mathrm{ms} \pm 12.1 \mathrm{μs}\left({\color{gray}0.028 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: medium, policies: 107	$$3.91 \mathrm{ms} \pm 19.2 \mathrm{μs}\left({\color{gray}-0.017 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: high, policies: 133	$$4.23 \mathrm{ms} \pm 24.7 \mathrm{μs}\left({\color{gray}2.75 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: low, policies: 1	$$3.26 \mathrm{ms} \pm 13.8 \mathrm{μs}\left({\color{gray}0.205 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: medium, policies: 63	$$3.79 \mathrm{ms} \pm 16.4 \mathrm{μs}\left({\color{gray}-1.846 \mathrm{\%}}\right) $$	Flame Graph

policy_resolution_none

Function	Value	Mean	Flame graphs
resolve_policies_for_actor	user: empty, selectivity: high, policies: 2	$$2.55 \mathrm{ms} \pm 8.90 \mathrm{μs}\left({\color{red}7.92 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: low, policies: 1	$$2.49 \mathrm{ms} \pm 11.3 \mathrm{μs}\left({\color{red}7.46 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: medium, policies: 1	$$2.60 \mathrm{ms} \pm 12.3 \mathrm{μs}\left({\color{red}7.98 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: high, policies: 8	$$2.75 \mathrm{ms} \pm 10.0 \mathrm{μs}\left({\color{red}5.33 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: low, policies: 1	$$2.67 \mathrm{ms} \pm 11.1 \mathrm{μs}\left({\color{red}6.36 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: medium, policies: 3	$$2.88 \mathrm{ms} \pm 15.3 \mathrm{μs}\left({\color{red}7.10 \mathrm{\%}}\right) $$	Flame Graph

policy_resolution_small

Function	Value	Mean	Flame graphs
resolve_policies_for_actor	user: empty, selectivity: high, policies: 52	$$2.91 \mathrm{ms} \pm 14.5 \mathrm{μs}\left({\color{red}6.59 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: low, policies: 1	$$2.62 \mathrm{ms} \pm 11.2 \mathrm{μs}\left({\color{red}8.71 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: empty, selectivity: medium, policies: 25	$$2.77 \mathrm{ms} \pm 11.9 \mathrm{μs}\left({\color{red}7.61 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: high, policies: 94	$$3.29 \mathrm{ms} \pm 20.2 \mathrm{μs}\left({\color{red}7.70 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: low, policies: 1	$$2.84 \mathrm{ms} \pm 11.9 \mathrm{μs}\left({\color{red}7.92 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: seeded, selectivity: medium, policies: 26	$$3.05 \mathrm{ms} \pm 17.5 \mathrm{μs}\left({\color{red}7.23 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: high, policies: 66	$$3.15 \mathrm{ms} \pm 15.3 \mathrm{μs}\left({\color{red}5.25 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: low, policies: 1	$$2.82 \mathrm{ms} \pm 14.2 \mathrm{μs}\left({\color{red}8.61 \mathrm{\%}}\right) $$	Flame Graph
resolve_policies_for_actor	user: system, selectivity: medium, policies: 29	$$3.08 \mathrm{ms} \pm 16.6 \mathrm{μs}\left({\color{red}8.39 \mathrm{\%}}\right) $$	Flame Graph

read_scaling_complete

Function	Value	Mean	Flame graphs
entity_by_id;one_depth	1 entities	$$38.8 \mathrm{ms} \pm 166 \mathrm{μs}\left({\color{gray}0.371 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;one_depth	10 entities	$$76.0 \mathrm{ms} \pm 397 \mathrm{μs}\left({\color{gray}-0.451 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;one_depth	25 entities	$$43.3 \mathrm{ms} \pm 178 \mathrm{μs}\left({\color{gray}-1.253 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;one_depth	5 entities	$$45.1 \mathrm{ms} \pm 167 \mathrm{μs}\left({\color{gray}-0.399 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;one_depth	50 entities	$$53.2 \mathrm{ms} \pm 246 \mathrm{μs}\left({\color{gray}-0.769 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;two_depth	1 entities	$$40.2 \mathrm{ms} \pm 144 \mathrm{μs}\left({\color{gray}-1.105 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;two_depth	10 entities	$$411 \mathrm{ms} \pm 910 \mathrm{μs}\left({\color{gray}-0.194 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;two_depth	25 entities	$$93.5 \mathrm{ms} \pm 391 \mathrm{μs}\left({\color{gray}0.415 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;two_depth	5 entities	$$84.3 \mathrm{ms} \pm 288 \mathrm{μs}\left({\color{gray}0.279 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;two_depth	50 entities	$$284 \mathrm{ms} \pm 664 \mathrm{μs}\left({\color{gray}3.48 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;zero_depth	1 entities	$$14.6 \mathrm{ms} \pm 65.3 \mathrm{μs}\left({\color{gray}2.04 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;zero_depth	10 entities	$$14.7 \mathrm{ms} \pm 70.5 \mathrm{μs}\left({\color{gray}0.876 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;zero_depth	25 entities	$$15.0 \mathrm{ms} \pm 61.1 \mathrm{μs}\left({\color{gray}0.811 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;zero_depth	5 entities	$$14.6 \mathrm{ms} \pm 62.2 \mathrm{μs}\left({\color{gray}0.452 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id;zero_depth	50 entities	$$17.4 \mathrm{ms} \pm 117 \mathrm{μs}\left({\color{gray}-1.426 \mathrm{\%}}\right) $$	Flame Graph

read_scaling_linkless

Function	Value	Mean	Flame graphs
entity_by_id	1 entities	$$14.5 \mathrm{ms} \pm 74.5 \mathrm{μs}\left({\color{gray}-1.137 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	10 entities	$$14.7 \mathrm{ms} \pm 74.7 \mathrm{μs}\left({\color{gray}0.228 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	100 entities	$$14.6 \mathrm{ms} \pm 66.4 \mathrm{μs}\left({\color{gray}1.80 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	1000 entities	$$15.1 \mathrm{ms} \pm 78.4 \mathrm{μs}\left({\color{gray}0.855 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	10000 entities	$$22.1 \mathrm{ms} \pm 130 \mathrm{μs}\left({\color{gray}-0.241 \mathrm{\%}}\right) $$	Flame Graph

representative_read_entity

Function	Value	Mean	Flame graphs
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/block/v/1`	$$29.1 \mathrm{ms} \pm 308 \mathrm{μs}\left({\color{gray}-1.239 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/book/v/1`	$$30.0 \mathrm{ms} \pm 205 \mathrm{μs}\left({\color{gray}4.21 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/building/v/1`	$$28.9 \mathrm{ms} \pm 231 \mathrm{μs}\left({\color{gray}-1.108 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/organization/v/1`	$$29.1 \mathrm{ms} \pm 281 \mathrm{μs}\left({\color{gray}-0.524 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/page/v/2`	$$29.1 \mathrm{ms} \pm 246 \mathrm{μs}\left({\color{gray}0.493 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/person/v/1`	$$29.1 \mathrm{ms} \pm 288 \mathrm{μs}\left({\color{gray}-4.814 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/playlist/v/1`	$$28.9 \mathrm{ms} \pm 257 \mathrm{μs}\left({\color{gray}-0.216 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/song/v/1`	$$29.1 \mathrm{ms} \pm 258 \mathrm{μs}\left({\color{gray}0.986 \mathrm{\%}}\right) $$	Flame Graph
entity_by_id	entity type ID: `https://blockprotocol.org/@alice/types/entity-type/uk-address/v/1`	$$29.2 \mathrm{ms} \pm 276 \mathrm{μs}\left({\color{gray}1.19 \mathrm{\%}}\right) $$	Flame Graph

representative_read_entity_type

Function	Value	Mean	Flame graphs
get_entity_type_by_id	Account ID: `bf5a9ef5-dc3b-43cf-a291-6210c0321eba`	$$8.04 \mathrm{ms} \pm 27.6 \mathrm{μs}\left({\color{gray}0.662 \mathrm{\%}}\right) $$	Flame Graph

representative_read_multiple_entities

Function	Value	Mean	Flame graphs
entity_by_property	traversal_paths=0	0	$$45.2 \mathrm{ms} \pm 293 \mathrm{μs}\left({\color{gray}-0.338 \mathrm{\%}}\right) $$
entity_by_property	traversal_paths=255	1,resolve_depths=inherit:1;values:255;properties:255;links:127;link_dests:126;type:true	$$92.7 \mathrm{ms} \pm 288 \mathrm{μs}\left({\color{gray}0.069 \mathrm{\%}}\right) $$
entity_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:0;links:0;link_dests:0;type:false	$$51.0 \mathrm{ms} \pm 380 \mathrm{μs}\left({\color{gray}-0.582 \mathrm{\%}}\right) $$
entity_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:0;links:1;link_dests:0;type:true	$$58.9 \mathrm{ms} \pm 281 \mathrm{μs}\left({\color{gray}-0.494 \mathrm{\%}}\right) $$
entity_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:2;links:1;link_dests:0;type:true	$$67.2 \mathrm{ms} \pm 326 \mathrm{μs}\left({\color{gray}-0.055 \mathrm{\%}}\right) $$
entity_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:2;properties:2;links:1;link_dests:0;type:true	$$74.2 \mathrm{ms} \pm 364 \mathrm{μs}\left({\color{gray}0.614 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=0	0	$$48.5 \mathrm{ms} \pm 198 \mathrm{μs}\left({\color{gray}-0.822 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=255	1,resolve_depths=inherit:1;values:255;properties:255;links:127;link_dests:126;type:true	$$75.1 \mathrm{ms} \pm 260 \mathrm{μs}\left({\color{gray}-0.892 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:0;links:0;link_dests:0;type:false	$$55.8 \mathrm{ms} \pm 357 \mathrm{μs}\left({\color{gray}0.457 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:0;links:1;link_dests:0;type:true	$$62.4 \mathrm{ms} \pm 253 \mathrm{μs}\left({\color{gray}-0.904 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:0;properties:2;links:1;link_dests:0;type:true	$$64.8 \mathrm{ms} \pm 311 \mathrm{μs}\left({\color{gray}-0.585 \mathrm{\%}}\right) $$
link_by_source_by_property	traversal_paths=2	1,resolve_depths=inherit:0;values:2;properties:2;links:1;link_dests:0;type:true	$$64.8 \mathrm{ms} \pm 320 \mathrm{μs}\left({\color{gray}0.091 \mathrm{\%}}\right) $$

scenarios

Function	Value	Mean	Flame graphs
full_test	query-limited	$$138 \mathrm{ms} \pm 451 \mathrm{μs}\left({\color{red}7.05 \mathrm{\%}}\right) $$	Flame Graph
full_test	query-unlimited	$$136 \mathrm{ms} \pm 535 \mathrm{μs}\left({\color{gray}3.71 \mathrm{\%}}\right) $$	Flame Graph
linked_queries	query-limited	$$39.3 \mathrm{ms} \pm 169 \mathrm{μs}\left({\color{lightgreen}-62.310 \mathrm{\%}}\right) $$	Flame Graph
linked_queries	query-unlimited	$$579 \mathrm{ms} \pm 888 \mathrm{μs}\left({\color{gray}-4.956 \mathrm{\%}}\right) $$	Flame Graph

vercel · 2026-01-15T13:31:49Z

The latest updates on your projects. Learn more about Vercel for GitHub.

2 Skipped Deployments

Project	Deployment	Review	Updated (UTC)
ds-theme	Ignored	Preview	Jan 15, 2026 3:23pm
hashdotdesign	Ignored	Preview	Jan 15, 2026 3:23pm

cursor · 2026-01-15T15:34:44Z

libs/@local/hashql/core/src/heap/mod.rs

+    unsafe fn rollback(&self, checkpoint: Self::Checkpoint) {
+        // SAFETY: Same safety guarantees as `Allocator::rollback`.
+        unsafe { self.inner.rollback(checkpoint) }
+    }


Heap::rollback doesn't clear interned strings, causing dangling pointers

High Severity

The new Heap::rollback implementation delegates directly to self.inner.rollback(checkpoint) without clearing the strings HashSet. The Heap struct stores interned strings as &'static str pointers that actually point into arena memory (the 'static lifetime is documented as "a lie"). When rollback resets the arena past where strings were interned, the strings HashSet retains dangling pointers. The reset() method correctly clears strings first with an explicit comment about preventing dangling references, but rollback() lacks this handling. This can cause use-after-free when checking string interning after a rollback.

I... actually didn't think about that. Good point the heap cannot be reset safely. I'll need to think about that.

indietyp added 3 commits January 1, 2026 18:02

chore: checkpoint

87377cc

feat: use heap for changed state tracking

f5bb523

chore: fix comment

70dc578

vercel bot temporarily deployed to Preview – petrinaut January 1, 2026 17:59 Inactive

graphite-app bot assigned indietyp Jan 1, 2026

github-actions bot added area/libs Relates to first-party libraries/crates/packages (area) type/eng > backend Owned by the @backend team labels Jan 1, 2026

indietyp mentioned this pull request Jan 1, 2026

BE-265: HashQL: Make CopyPropagation aware of local re-assignments #8231

Open

3 tasks

indietyp force-pushed the bm/be-267-hashql-do-not-reset-the-scratch-space-but-instead-use-scopes branch from 47b5f9f to 67bcdb0 Compare January 1, 2026 18:00

vercel bot temporarily deployed to Preview – petrinaut January 1, 2026 18:00 Inactive

fix: import

2c43268

vercel bot temporarily deployed to Preview – petrinaut January 1, 2026 18:03 Inactive

augmentcode bot reviewed Jan 1, 2026

View reviewed changes

libs/@local/hashql/mir/benches/transform.rs Outdated Show resolved Hide resolved

vercel bot deployed to Preview – hash January 1, 2026 18:10 View deployment

chore: remove dead reference

3bcc7e9

vercel bot temporarily deployed to Preview – petrinaut January 2, 2026 15:12 Inactive

cursor bot reviewed Jan 2, 2026

View reviewed changes

libs/@local/hashql/mir/benches/transform.rs Outdated Show resolved Hide resolved

libs/@local/hashql/mir/benches/transform.rs Outdated Show resolved Hide resolved

graphite-app bot requested review from a team January 2, 2026 15:26

fix: appease the AI overlords

cb4fa90

vercel bot temporarily deployed to Preview – petrinaut January 2, 2026 15:49 Inactive

This was referenced Jan 10, 2026

BE-272: HashQL: Add numeric addition and subtraction operators to MIR #8250

Open

BE-273: HashQL: Profile and benchmark MIR interpreter #8251

Open

indietyp force-pushed the bm/be-266-hashql-implement-pre-inlining-fix-point-loop branch from 85aff24 to 99146b5 Compare January 15, 2026 13:31

indietyp force-pushed the bm/be-267-hashql-do-not-reset-the-scratch-space-but-instead-use-scopes branch from 2dcec85 to 789693c Compare January 15, 2026 13:31

vercel bot deployed to Preview – petrinaut January 15, 2026 13:37 View deployment

vercel bot deployed to Preview – hash January 15, 2026 13:45 View deployment

indietyp force-pushed the bm/be-267-hashql-do-not-reset-the-scratch-space-but-instead-use-scopes branch from 789693c to cb4fa90 Compare January 15, 2026 15:22

indietyp force-pushed the bm/be-266-hashql-implement-pre-inlining-fix-point-loop branch from 99146b5 to 720ee51 Compare January 15, 2026 15:22

vercel bot deployed to Preview – petrinaut January 15, 2026 15:27 View deployment

vercel bot deployed to Preview – hash January 15, 2026 15:34 View deployment

cursor bot reviewed Jan 15, 2026

View reviewed changes

BE-267: Use bump scopes instead of full resets across transformation passes #8234

Are you sure you want to change the base?

BE-267: Use bump scopes instead of full resets across transformation passes #8234

Uh oh!

Conversation

indietyp commented Jan 1, 2026

🌟 What is the purpose of this PR?

🔍 What does this change?

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

📜 Does this require a change to the docs?

🕸️ Does this require a change to the Turbo Graph?

🛡 What tests cover this?

❓ How to test this?

Uh oh!

cursor bot commented Jan 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Summary

Uh oh!

indietyp commented Jan 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

augmentcode bot commented Jan 1, 2026

Uh oh!

augmentcode bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

codspeed-hq bot commented Jan 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CodSpeed Performance Report

Merging this PR will not alter performance

Summary

Footnotes

Uh oh!

codecov bot commented Jan 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Uh oh!

Uh oh!

github-actions bot commented Jan 2, 2026

Benchmark results

@rust/hash-graph-benches – Integrations

policy_resolution_large

policy_resolution_medium

policy_resolution_none

policy_resolution_small

read_scaling_complete

read_scaling_linkless

representative_read_entity

representative_read_entity_type

representative_read_multiple_entities

scenarios

Uh oh!

vercel bot commented Jan 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cursor bot Jan 15, 2026

Choose a reason for hiding this comment

Heap::rollback doesn't clear interned strings, causing dangling pointers

Uh oh!

indietyp Jan 15, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

2 participants

cursor bot commented Jan 1, 2026 •

edited

Loading

indietyp commented Jan 1, 2026 •

edited

Loading

codspeed-hq bot commented Jan 1, 2026 •

edited

Loading

codecov bot commented Jan 1, 2026 •

edited

Loading

vercel bot commented Jan 15, 2026 •

edited

Loading

`Heap::rollback` doesn't clear interned strings, causing dangling pointers