feat: ReplaceTypes: handlers for array constants + linearization #2023

acl-cqc · 2025-03-25T12:28:14Z

Final chapter closes #1948.

default() includes all the handlers, new_empty() does not.

… loops

…lmost all cases

…pes.rs

…d Const.

codecov · 2025-04-08T20:14:36Z

Codecov Report

Attention: Patch coverage is 95.92593% with 11 lines in your changes missing coverage. Please review.

Project coverage is 83.05%. Comparing base (9ad9e6d) to head (63dc9a1).
Report is 13 commits behind head on main.

Files with missing lines	Patch %	Lines
hugr-passes/src/replace_types/handlers.rs	95.43%	4 Missing and 6 partials ⚠️
hugr-passes/src/replace_types.rs	96.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2023      +/-   ##
==========================================
- Coverage   83.31%   83.05%   -0.26%     
==========================================
  Files         217      218       +1     
  Lines       41864    41783      -81     
  Branches    38078    37961     -117     
==========================================
- Hits        34877    34703     -174     
- Misses       5033     5275     +242     
+ Partials     1954     1805     -149

Flag	Coverage Δ
python	`85.40% <ø> (-0.53%)`	⬇️
rust	`82.81% <95.92%> (-0.24%)`	⬇️

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.

mark-koch · 2025-04-09T08:41:49Z

hugr-passes/src/replace_types/handlers.rs

+    // If none of the values has changed, assume the Type hasn't (Values have a single known type)
+    if !ch {
+        return Ok(None);
+    };


Doesn't this break for empty lists?

Yes, good spot.
list_const was originally more for checking the framework did the right thing(s) for constants, rather than being a definitive handler for lists, but given it was pub it should have been better ;-). I've fixed this, although I haven't added a test with an empty list - the array_const test does include such a case and the SUT is identical.... (I did try commoning list_const and array_const up but it needed 3 different impl Fns so was not worth it!)

mark-koch · 2025-04-09T09:03:03Z

hugr-passes/src/replace_types/linearize.rs

+    #[rstest]
+    fn array(#[values(2, 3, 4)] num_outports: usize) {


Are lists constants already tested somewhere?

The loop_const does some fairly funky things with them, although is not exhaustive

mark-koch · 2025-04-09T09:04:53Z

hugr-passes/src/replace_types/handlers.rs

@@ -0,0 +1,240 @@
+//! Callbacks for use with [ReplaceTypes::replace_consts_parametrized]
+//! and [ReplaceTypes::linearize_parametric]


I think this reference breaks the docs build?

mark-koch · 2025-04-09T09:05:06Z

hugr-passes/src/replace_types/handlers.rs

+    h.signature(h.root()).unwrap().runtime_reqs.clone()
+}
+
+/// Handler for copying/discarding arrays, for use with [ReplaceTypes::linearize_parametric] of


Also this one

hugr-passes/src/replace_types/handlers.rs

acl-cqc · 2025-04-15T12:07:01Z

@mark-koch added an array_const handler, fixed the bug in list_const (looks just like array, tho that case isn't tested), added a ConstTypeError variant to ReplaceTypesError (thankfully non-breaking as RTErr is non-exhaustive). Can you take another look?

Any thoughts on what to do with the handlers? I do feel including them in ReplaceTypes::default is tempting, although indeed I'm not keen on "remove handlers if you don't want them". (Why would one not want them, though - basically that's saying, you just want to break sooner and harder if you find a Hugr with lists/arrays...) Could be another method, although in which case naming is an issue (default with handlers and minimal without? default and new_with_collections? I wouldn't want to imply new_with_collections provided a complete set of handlers...)

mark-koch

Thanks, looks good 👍

Any thoughts on what to do with the handlers?

I think handlers for std extensions should be included in default(). Maybe ReplaceTypes::empty() to get one without any handlers?

acl-cqc · 2025-04-15T17:38:26Z

Thanks, looks good 👍

Any thoughts on what to do with the handlers?

I think handlers for std extensions should be included in default(). Maybe ReplaceTypes::empty() to get one without any handlers?

That was not quite as neat as I'd have liked, 'coz we can't get the Default-derive macro to do the work, but done.

EDIT: so, hmmm, I am considering pub fn with_std(&mut self) -> () or pub fn with_std(self) -> Self

## 🤖 New release * `hugr-model`: 0.19.0 -> 0.19.1 (✓ API compatible changes) * `hugr-core`: 0.15.3 -> 0.15.4 (✓ API compatible changes) * `hugr-llvm`: 0.15.3 -> 0.15.4 (~⚠ API breaking changes~ overwriten) * `hugr-passes`: 0.15.3 -> 0.15.4 (✓ API compatible changes) * `hugr`: 0.15.3 -> 0.15.4 (✓ API compatible changes) * `hugr-cli`: 0.15.3 -> 0.15.4 (✓ API compatible changes) ### ⚠ `hugr-llvm` breaking changes ```text --- failure trait_missing: pub trait removed or renamed --- Description: A publicly-visible trait cannot be imported by its prior path. A `pub use` may have been removed, or the trait itself may have been renamed or removed entirely. ref: https://doc.rust-lang.org/cargo/reference/semver.html#item-remove impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.40.0/src/lints/trait_missing.ron Failed in: trait hugr_llvm::utils::array_op_builder::ArrayOpBuilder, previously in file /tmp/.tmpyTFqSG/hugr-llvm/src/utils/array_op_builder.rs:10 trait hugr_llvm::utils::ArrayOpBuilder, previously in file /tmp/.tmpyTFqSG/hugr-llvm/src/utils/array_op_builder.rs:10 ``` <details><summary>Changelog</summary> ## `hugr-model` <blockquote> ## [0.19.0](hugr-model-v0.18.1...hugr-model-v0.19.0) - 2025-04-02 ### New Features - Python bindings for `hugr-model`. ([#1959](#1959)) - Remove extension sets from `hugr-model`. ([#2031](#2031)) - Packages in `hugr-model` and envelope support. ([#2026](#2026)) - Represent order edges in `hugr-model` as metadata. ([#2027](#2027)) </blockquote> ## `hugr-core` <blockquote> ## [0.16.0](hugr-core-v0.15.3...hugr-core-v0.16.0) - 2025-04-30 ### New Features - Export the portgraph hierarchy in HugrInternals ([#2057](#2057)) - Implement Debug for generic Wire<N>s ([#2068](#2068)) - Add ExtensionOp helpers ([#2072](#2072)) - ReplaceTypes: handlers for array constants + linearization ([#2023](#2023)) - move `ArrayOpBuilder` to hugr-core ([#2115](#2115)) ### Testing - Disable IO-dependent tests when running miri ([#2123](#2123)) </blockquote> ## `hugr-llvm` <blockquote> ## [0.16.0](hugr-llvm-v0.15.3...hugr-llvm-v0.16.0) - 2025-04-30 ### New Features - move `ArrayOpBuilder` to hugr-core ([#2115](#2115)) </blockquote> ## `hugr-passes` <blockquote> ## [0.16.0](hugr-passes-v0.15.3...hugr-passes-v0.16.0) - 2025-04-30 ### New Features - ReplaceTypes: handlers for array constants + linearization ([#2023](#2023)) </blockquote> ## `hugr` <blockquote> ## [0.16.0](hugr-v0.15.3...hugr-v0.16.0) - 2025-04-30 ### New Features - Export the portgraph hierarchy in HugrInternals ([#2057](#2057)) - Implement Debug for generic Wire<N>s ([#2068](#2068)) - Add ExtensionOp helpers ([#2072](#2072)) - ReplaceTypes: handlers for array constants + linearization ([#2023](#2023)) - move `ArrayOpBuilder` to hugr-core ([#2115](#2115)) ### Testing - Disable IO-dependent tests when running miri ([#2123](#2123)) </blockquote> ## `hugr-cli` <blockquote> ## [0.15.3](hugr-cli-v0.15.2...hugr-cli-v0.15.3) - 2025-04-02 ### Documentation - Add usage info to hugr-cli's rustdocs ([#2044](#2044)) </blockquote> </details> --- This PR was generated with [release-plz](https://github.com/release-plz/release-plz/). --------- Co-authored-by: Agustín Borgna <agustin.borgna@quantinuum.com>

acl-cqc added 30 commits March 17, 2025 19:14

Add TypeTransformer and (Sum/Function)Type(Arg/Row)::transform

937cd14

trait Transformable to common up all the 'let mut any_change = false'…

815536b

… loops

Transformable v2: implement for [E], works without iter_mut/etc. in a…

cc1e8f2

…lmost all cases

first test, fix CustomType bound caching

4888184

Second test, fix SumType caching

7002e4d

Make clippy happy

6cc7cea

Add HugrMut::optype_mut (v2, allow mutating root if RootHandle == Node)

d43784b

WIP add hugr-passes/src/lower_types.rs (w/ change_node, subst_ty)

422c496

Add def_arc

76ed391

change_node, change_type, and the rest

ed5b5dd

Use TypeTransformer framework, removing most type_stuff from lower_ty…

adcbbf6

…pes.rs

Add a load of copy/discard lowering stuff, OpReplacement

d84ae4e

OpHashWrapper

bfa52cf

Parametrized type support

b373571

remove copy_discard stuff

a8e613a

Assume less in OpReplacement::replace

24ed15c

parametrized ops

9153fcf

Comments, renaming, use const_fn

a89879c

Comment const_fn TODO

c78d88a

Test panics on unexpected argument - simpler, better

bf6a9a4

test functiontype

5ecd9e6

clippy that new test

d9a6d29

Merge remote-tracking branch 'origin/main' into acl/type_transform

83e798e

WIP setup for test

84fe82d

First test

6168353

read only makes sense for Copyables

fcb85a2

Extend test to Calls of polyfunc; comments, monomorphize first

0f9aa17

no, instantiate the calls with types being lowered

74c6775

Consts: HashMap keyed by either, add lower_ methods. Test TailLoop an…

00fd284

…d Const.

clippy, turn off type-complexity

9f02acf

acl-cqc requested a review from a team as a code owner April 8, 2025 20:12

acl-cqc requested review from jake-arkinstall and mark-koch and removed request for jake-arkinstall April 8, 2025 20:12

mark-koch approved these changes Apr 9, 2025

View reviewed changes

mark-koch reviewed Apr 9, 2025

View reviewed changes

hugr-passes/src/replace_types/handlers.rs Show resolved Hide resolved

fix docs

318c7fc

doug-q added this to the hugr-rs 0.16 milestone Apr 15, 2025

ss2165 modified the milestones: hugr-rs 0.16, hugr-rs 0.15.x Apr 15, 2025

acl-cqc added 2 commits April 15, 2025 12:00

Fix list_const, add array_const

f5de2ec

test, add ReplaceTypesError::ConstError

6ae093a

acl-cqc requested a review from mark-koch April 15, 2025 12:02

acl-cqc changed the title ~~feat: Provide ReplaceTypes handlers for linearizing arrays~~ feat:ReplaceTypes: handlers for array constants + linearization Apr 15, 2025

acl-cqc changed the title ~~feat:ReplaceTypes: handlers for array constants + linearization~~ feat: ReplaceTypes: handlers for array constants + linearization Apr 15, 2025

mark-koch approved these changes Apr 15, 2025

View reviewed changes

acl-cqc added 2 commits April 15, 2025 18:33

Default includes handlers, new_empty does not

ff1772f

docs

63dc9a1

acl-cqc mentioned this pull request Apr 15, 2025

Reduce duplication when linearizing arrays #2086

Open

acl-cqc enabled auto-merge April 16, 2025 10:18

acl-cqc added this pull request to the merge queue Apr 16, 2025

Merged via the queue into main with commit a4f5196 Apr 16, 2025
27 checks passed

acl-cqc deleted the acl/lower_types_linearize_array branch April 16, 2025 10:22

hugrbot mentioned this pull request Apr 16, 2025

chore: release v0.19.1 #2079

Closed

hugrbot mentioned this pull request Apr 28, 2025

chore: release #2121

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat: ReplaceTypes: handlers for array constants + linearization #2023

feat: ReplaceTypes: handlers for array constants + linearization #2023

Uh oh!

acl-cqc commented Mar 25, 2025 •

edited

Loading

Uh oh!

codecov bot commented Apr 8, 2025 •

edited

Loading

Uh oh!

mark-koch Apr 9, 2025

Uh oh!

acl-cqc Apr 15, 2025

Uh oh!

mark-koch Apr 9, 2025

Uh oh!

acl-cqc Apr 15, 2025

Uh oh!

mark-koch Apr 9, 2025

Uh oh!

mark-koch Apr 9, 2025

Uh oh!

Uh oh!

acl-cqc commented Apr 15, 2025

Uh oh!

mark-koch left a comment

Uh oh!

acl-cqc commented Apr 15, 2025 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

		@@ -0,0 +1,240 @@
		//! Callbacks for use with [ReplaceTypes::replace_consts_parametrized]
		//! and [ReplaceTypes::linearize_parametric]

feat: ReplaceTypes: handlers for array constants + linearization #2023

feat: ReplaceTypes: handlers for array constants + linearization #2023

Uh oh!

Conversation

acl-cqc commented Mar 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Apr 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

mark-koch Apr 9, 2025

Choose a reason for hiding this comment

Uh oh!

acl-cqc Apr 15, 2025

Choose a reason for hiding this comment

Uh oh!

mark-koch Apr 9, 2025

Choose a reason for hiding this comment

Uh oh!

acl-cqc Apr 15, 2025

Choose a reason for hiding this comment

Uh oh!

mark-koch Apr 9, 2025

Choose a reason for hiding this comment

Uh oh!

mark-koch Apr 9, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

acl-cqc commented Apr 15, 2025

Uh oh!

mark-koch left a comment

Choose a reason for hiding this comment

Uh oh!

acl-cqc commented Apr 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

acl-cqc commented Mar 25, 2025 •

edited

Loading

codecov bot commented Apr 8, 2025 •

edited

Loading

acl-cqc commented Apr 15, 2025 •

edited

Loading