New Reference Logic by raphael-goetz · Pull Request #121 · code0-tech/taurus

raphael-goetz · 2026-03-16T14:23:10Z

Resolves: #97, #98, #113, #117

…uite)

…e error instead of panicing

…nto major-rework

…rate was referncing core lib

…ctions

Copilot

Pull request overview

This PR refactors the Taurus execution engine into a new taurus-core crate, adds execution tracing and new reference-resolution behavior, and introduces a JSON-driven execution test suite that runs in CI.

Changes:

Extract core execution/context/runtime logic into crates/core (taurus-core) and update the Taurus service to consume it.
Add execution tracing infrastructure (trace frames + renderer) and expand reference handling (flow input, inputType, reference paths).
Add an “execution suite” binary (crates/tests) with example flows and wire it into the GitHub Actions workflow.

Reviewed changes

Copilot reviewed 33 out of 38 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
taurus/src/context/executor.rs	Removed old executor implementation (superseded by `taurus-core`).
taurus/src/context/context.rs	Removed old context implementation (superseded by `taurus-core`).
crates/core/src/lib.rs	New library root exporting `context`, `debug`, and `runtime`.
crates/core/src/context/mod.rs	Context module wiring for the new core crate.
crates/core/src/context/argument.rs	Updated runtime error import path for arguments.
crates/core/src/context/context.rs	New context with flow input, inputType storage, and reference path traversal.
crates/core/src/context/executor.rs	New executor with eager/lazy thunk evaluation and optional trace rendering.
crates/core/src/context/macros.rs	Updated macros to use `runtime::error::RuntimeError`.
crates/core/src/context/registry.rs	`FunctionStore::default()` now auto-populates all runtime functions.
crates/core/src/context/signal.rs	Updated signal to use the new runtime error type.
crates/core/src/debug/mod.rs	New debug module exports.
crates/core/src/debug/trace.rs	Data model for execution trace frames/edges/outcomes.
crates/core/src/debug/tracer.rs	Trace collector implementation (`Tracer`).
crates/core/src/debug/render.rs	Human-readable trace renderer.
crates/core/src/runtime/mod.rs	Runtime module root.
crates/core/src/runtime/error/mod.rs	Runtime error type now exposes `name` and `message` publicly.
crates/core/src/runtime/functions/mod.rs	Central function collection used to populate the registry.
crates/core/src/runtime/functions/array.rs	Updated list APIs to support lazy predicate/transform thunks and inputType iteration.
crates/core/src/runtime/functions/boolean.rs	Updated imports to use new runtime error module.
crates/core/src/runtime/functions/control.rs	Adds runtime trace labels for branching.
crates/core/src/runtime/functions/http.rs	Updates headers handling from list to struct.
crates/core/src/runtime/functions/number.rs	Fixes random range panic and adds tests for invalid ranges.
crates/core/src/runtime/functions/object.rs	New object helper functions (+ unit tests).
crates/core/src/runtime/functions/text.rs	Updated imports to use new runtime error module.
crates/core/Cargo.toml	Introduces `taurus-core` crate and its dependencies.
crates/core/.gitignore	Ignores core crate build output.
crates/taurus/src/config/mod.rs	Adds a configuration module for the Taurus service.
crates/taurus/src/main.rs	Switches Taurus service to use `taurus-core` executor/registry and enables tracing.
crates/taurus/Cargo.toml	Moves Taurus crate to workspace version/edition and depends on `taurus-core`.
crates/tests/Cargo.toml	Adds a `tests` binary crate for the execution suite.
crates/tests/src/main.rs	JSON-driven execution suite runner for validation flows.
crates/tests/README.md	Documentation for running/adding flows to the suite.
crates/tests/flows/01_return_object.json	Example flow file (response object).
crates/tests/flows/02_return_flow_input.json	Example flow file (flow input reference).
crates/tests/flows/03_for_each.json	Example flow file (inputType reference via `for_each`).
Cargo.toml	Expands workspace members and adds workspace deps (incl. `taurus-core`).
Cargo.lock	Dependency lockfile updates (workspace + tucana bump).
.github/workflows/build-and-test.yml	Runs both `cargo test` and the new execution suite in CI.

Comments suppressed due to low confidence (8)

crates/core/src/runtime/functions/array.rs:188

filter inserts an InputType into the Context each iteration, but several early-return paths don't clear it (e.g., Err(e) => return Signal::Failure(e) and returning other signals). This leaves stale input_types entries that can affect subsequent executions. Consider clearing input_type on all exit paths (RAII guard / defer-style helper).
crates/core/src/runtime/functions/array.rs:453
for_each has two issues here: (1) the error message says "map expects..." even though this is for_each, and (2) if the thunk returns a non-Success signal, the function returns early without calling ctx.clear_input_type(input_type), leaving stale context state. Consider fixing the message and ensuring cleanup happens before any early return.
crates/core/src/runtime/functions/array.rs:724
sort populates two InputType entries (input_index 0 and 1) but never clears them on the success path (and only clears input_type, not input_type_next, on error paths). This can leak stale values into later reference lookups. Ensure both input types are cleared before returning, regardless of success/failure.
crates/core/src/runtime/functions/number.rs:389
The random_number function registers as std::number::random_number, but this error message mentions std::math::random and has a grammar typo ("then" -> "than"). Consider updating the message to reference the correct function name and fix the wording to make debugging clearer.
crates/taurus/src/main.rs:31
handle_message always enables tracing (with_trace = true), which will print a potentially large per-execution trace to stdout for every flow. This can significantly increase log volume and CPU in production. Consider gating tracing behind a config/env flag (e.g., only in development) or using structured logging instead of unconditional println!.
crates/core/src/runtime/functions/array.rs:740
sort runs the comparator thunk inside a sort_by closure but always returns Ordering::Equal, then performs a second sort_by using a sequential out vector unrelated to the compared elements. This violates sort_by's comparator requirements (total, consistent ordering) and can yield incorrect / non-deterministic results. Consider computing the comparator result for the actual (a,b) pair and returning Less/Equal/Greater directly from the first sort_by (capturing failures to return after the sort), rather than sorting twice.
crates/core/src/runtime/functions/array.rs:834
sort_reverse has the same issue as sort: the first sort_by executes the comparator thunk but always returns Ordering::Equal, and the second sort_by uses a sequential out vector that doesn't correspond to the compared elements. This can produce incorrect / non-deterministic ordering and violates sort_by comparator expectations. Consider returning the ordering derived from the thunk directly within the comparator closure (capturing failures to handle after sorting).
crates/core/src/runtime/functions/array.rs:818
sort_reverse also never clears input_type/input_type_next on the success path, and on error paths only clears input_type. This can leak stale comparison operands into later inputType references. Ensure both entries are cleared before returning, regardless of outcome.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

crates/core/src/context/context.rs

crates/tests/flows/01_return_object.json

crates/core/src/context/executor.rs

crates/tests/src/main.rs

crates/tests/flows/03_for_each.json

crates/tests/README.md

crates/core/src/context/executor.rs

crates/core/src/context/context.rs

crates/tests/src/main.rs

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> Signed-off-by: Raphael Götz <52959657+raphael-goetz@users.noreply.github.com>

Copilot

Pull request overview

This PR restructures Taurus into a workspace with a new taurus-core crate, adds an execution test-suite crate with exported flow fixtures, and introduces new reference/tracing logic in the executor and runtime functions (including support for new reference value targets like flow input and input-type references).

Changes:

Introduces taurus-core (context/executor/runtime/debug modules) and migrates Taurus to use it.
Adds a standalone execution suite (crates/tests) that runs exported flow fixtures in CI.
Updates runtime functions to support new reference behaviors (e.g., InputType) and improves debugging/tracing; fixes invalid random ranges.

Reviewed changes

Copilot reviewed 33 out of 38 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
taurus/src/context/executor.rs	Removed old executor implementation (migrated into `taurus-core`).
taurus/src/context/context.rs	Removed old context implementation (migrated into `taurus-core`).
crates/tests/src/main.rs	Adds execution-suite runner for JSON flow fixtures.
crates/tests/flows/03_for_each.json	Adds a flow fixture exercising `for_each` + `inputType` references.
crates/tests/flows/02_return_flow_input.json	Adds a flow fixture exercising `flowInput` references.
crates/tests/flows/01_return_object.json	Adds a flow fixture for a simple response-returning chain.
crates/tests/README.md	Documents how to run/extend the execution-suite.
crates/tests/Cargo.toml	Adds `tests` crate dependencies and workspace integration.
crates/taurus/src/main.rs	Switches Taurus binary to `taurus-core` and enables tracing in execution.
crates/taurus/src/config/mod.rs	Introduces a config module for Taurus startup configuration.
crates/taurus/Cargo.toml	Uses workspace version/edition and adds dependency on `taurus-core`.
crates/core/src/runtime/mod.rs	Adds runtime module boundary for functions and errors.
crates/core/src/runtime/functions/text.rs	Updates imports to use `runtime::error::RuntimeError`.
crates/core/src/runtime/functions/object.rs	Adds new object stdlib functions + unit tests.
crates/core/src/runtime/functions/number.rs	Updates error import; changes random range handling + adds tests.
crates/core/src/runtime/functions/mod.rs	Adds runtime function collection entry-point.
crates/core/src/runtime/functions/http.rs	Changes HTTP headers representation to `Struct` and updates validation.
crates/core/src/runtime/functions/control.rs	Adds runtime trace labels for branch selection.
crates/core/src/runtime/functions/boolean.rs	Updates imports to use `runtime::error::RuntimeError`.
crates/core/src/runtime/functions/array.rs	Adds lazy thunk support for list ops; adds `InputType` injection and tracing.
crates/core/src/runtime/error/mod.rs	Makes `RuntimeError.name/message` public for suite output.
crates/core/src/lib.rs	Exposes `context`, `debug`, and `runtime` modules.
crates/core/src/debug/tracer.rs	Adds tracer implementation collecting execution frames.
crates/core/src/debug/trace.rs	Adds trace data model (frames/edges/outcomes).
crates/core/src/debug/render.rs	Adds ANSI tree renderer for execution traces.
crates/core/src/debug/mod.rs	Exposes debug submodules.
crates/core/src/context/signal.rs	Switches to `runtime::error::RuntimeError`.
crates/core/src/context/registry.rs	`FunctionStore::default()` now populates via `runtime::functions::collect()`.
crates/core/src/context/mod.rs	Adds `context` module re-exports.
crates/core/src/context/macros.rs	Updates macros to use `runtime::error::RuntimeError`.
crates/core/src/context/executor.rs	New executor: reference resolution, thunk handling, and optional trace generation.
crates/core/src/context/context.rs	New context: stores results, flow input, input types, and supports reference paths.
crates/core/src/context/argument.rs	Switches to `runtime::error::RuntimeError`.
crates/core/Cargo.toml	Introduces `taurus-core` crate manifest.
crates/core/.gitignore	Ignores crate-local build artifacts.
Cargo.toml	Converts repo into workspace; bumps `tucana` and adds shared deps.
Cargo.lock	Updates lockfile for new workspace crates and dependency bump.
.github/workflows/build-and-test.yml	Runs `cargo test` and the new execution-suite in CI.

Comments suppressed due to low confidence (12)

crates/core/src/runtime/functions/array.rs:409

The argument-shape error message in for_each says map expects ..., which is misleading when debugging failures. Update the message to reference for_each and its expected arg names/modes.
crates/core/src/runtime/functions/array.rs:184
std::list::filter sets an InputType entry in the context, but several early-return paths skip ctx.clear_input_type(input_type) (e.g., when the predicate returns non-bool or when run returns Failure/Return/Respond/Stop). This can leave stale input_type values in the context for subsequent executions.
crates/core/src/runtime/functions/array.rs:716
sort executes the comparator thunk inside a sort_by closure but always returns Ordering::Equal. The standard library is free to call the comparator any number of times and in any order, so the collected signals/out sequence will be nondeterministic and may not match the comparisons performed by the second sort_by, producing incorrect ordering. Consider performing a single sort where each comparison calls the comparator thunk and returns Ordering directly (and propagate errors deterministically, e.g., by using a custom stable sort implementation that can short-circuit on failure).
crates/core/src/runtime/functions/array.rs:679
The argument-shape error message in sort says map expects ..., which is misleading when debugging. Update the message to reference sort and the comparator arg modes (array: eager, comparator: lazy thunk).
crates/core/src/runtime/functions/number.rs:396
rand::random_range(min..=max) uses an inclusive range, but the existing test asserts the result is < max (and callers often expect a half-open range for floats). This can make tests flaky (e.g., returning exactly max) and may be a behavior change from the previous integer-based implementation. Consider using a half-open range (min..max) for floats (handling min==max separately), or update tests/contract to allow <= max.
crates/core/src/runtime/functions/array.rs:433
std::list::for_each returns early on Failure/Return/Respond/Stop without clearing the inserted InputType entry, so the context can retain stale input_type values. Clear the input_type on all exit paths (e.g., via a scope guard / defer-style helper).
crates/core/src/runtime/functions/number.rs:389
The random error message has a few issues: grammar uses "then" instead of "than", and it references std::math::random even though this handler is registered as std::number::random_number. Consider correcting the wording to avoid confusing users.
crates/core/src/runtime/functions/number.rs:811
Test name has typos: test_random_range_fist_bigger_then_second should be ...first_bigger_than_second ("fist"/"then"). Renaming improves readability and searchability.
crates/core/src/runtime/functions/array.rs:810
sort_reverse has the same nondeterminism issue as sort: it runs the comparator thunk inside a sort_by that always returns Ordering::Equal, so the number/order of thunk executions (and thus out) is unspecified and can lead to incorrect ordering in the subsequent sort. Consider a single deterministic comparison path (custom sort) that calls the comparator and returns Ordering directly.
crates/core/src/runtime/functions/array.rs:740
sort and sort_reverse insert two InputType entries (input_index 0 and 1), but cleanup only ever clears input_type (index 0) and never clears input_type_next (index 1), and there is no cleanup on the success path. This will leave stale input_type values in Context after sorting.
crates/taurus/src/main.rs:31
handle_message always enables tracing (with_trace=true), which will print a full execution trace for every flow execution. This is likely too noisy/expensive for normal operation; wire this to configuration (env var / log level) so trace rendering is only enabled when debugging.
crates/core/src/runtime/functions/array.rs:773
The argument-shape error message in sort_reverse says map expects ..., which is misleading. Update it to reference sort_reverse and its expected args (array eager, comparator lazy thunk).

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

crates/core/src/context/context.rs

+        }
+    }
+
+    pub fn get_current_node_id(&mut self) -> i64 {


crates/tests/src/main.rs

+    pub fn run_tests(&self) {
+        for case in self.cases.clone() {
+            match case.run() {
+                CaseResult::Success => print_success(&case),
+                CaseResult::Failure(input, result) => print_failure(&case, &input, result),
+            }
+        }
+    }


crates/core/src/context/executor.rs

+    pub fn execute(&self, start_node_id: i64, ctx: &mut Context, with_trace: bool) -> Signal {
+        let mut tracer = Tracer::new();
+
+        let (signal, _root_frame) = self.execute_call(start_node_id, ctx, &mut tracer);
+
+        if with_trace && let Some(run) = tracer.take_run() {
+            println!("{}", crate::debug::render::render_trace(&run));
+        }


crates/tests/flows/02_return_flow_input.json

+{
+	"name": "02_return_object",
+	"description": "This flow expects the same output value as the input",


crates/core/src/context/executor.rs

+        let node = match self.nodes.get(&node_id) {
+            Some(n) => n.clone(),
+            None => {
+                let err =
+                    RuntimeError::simple("NodeNotFound", format!("Node {} not found", node_id));
+                return (Signal::Failure(err), 0);
+            }
+        };
+
+        let entry = match self.functions.get(node.runtime_function_id.as_str()) {
+            Some(e) => e,
+            None => {
+                let err = RuntimeError::simple(
+                    "FunctionNotFound",
+                    format!("Function {} not found", node.runtime_function_id),
+                );
+                return (Signal::Failure(err), 0);
+            }


crates/core/src/context/context.rs

+                                if let Kind::StructValue(struct_value) = &kind {
+                                    match struct_value.fields.get(part) {
+                                        Some(x) => {
+                                            curr = x.clone();
+                                        }
+                                        None => return ContextResult::NotFound,
+                                    }


crates/tests/src/main.rs

+                taurus_core::context::signal::Signal::Success(value) => {
+                    let json = to_json_value(value);
+                    if json == input.clone().expected_result {
+                        return CaseResult::Success;
+                    } else {
+                        return CaseResult::Failure(input, json);
+                    }
+                }
+                taurus_core::context::signal::Signal::Return(value) => {
+                    let json = to_json_value(value);
+                    if json == input.clone().expected_result {
+                        return CaseResult::Success;
+                    } else {
+                        return CaseResult::Failure(input, json);
+                    }
+                }
+                taurus_core::context::signal::Signal::Respond(value) => {
+                    let json = to_json_value(value);
+                    if json == input.clone().expected_result {
+                        return CaseResult::Success;
+                    } else {
+                        return CaseResult::Failure(input, json);
+                    }
+                }


crates/tests/flows/03_for_each.json

+	"name": "03_for_each",
+	"description": "This flow validates the functionality of the refernece type 'input_type'",
+	"inputs": [


crates/tests/flows/01_return_object.json

+	"flow": {
+		"starting_node_id": "1",
+		"node_functions": [
+			{


crates/core/src/debug/render.rs

+        ));
+
+        let edge_child_prefix = format!("{}{}", prefix, if last { "   " } else { "│  " });
+        render_frame(*child_id, by_id, &edge_child_prefix, true, out);


raphael-goetz added 30 commits February 23, 2026 13:18

feat: reworked execution logic

59fea78

feat: new reference logic impl

f759cef

feat: added tracing

da492a9

feat: wip new implementation of array functions

6165a14

dependencies: updated to tucana 0.0.53

eaa67c8

feat: added new module

e9ce54c

feat: added current node id to context

5925b4c

feat: set current node id in execution loop

c4083e1

feat: added new implementation for iterators needing input types

9e75568

dependencies: updated tucana and code0-flow

6628165

ref: cargo fmt

2eb7899

ref: cargo clippy

1747cf2

ref: split taurus codebase into core, taurus (service), tests (test-s…

4d3a7bb

…uite)

fix: random_range if numbers are in wrong order it will return runtim…

6b648fc

…e error instead of panicing

feat: added wip flow test execution suite

baacba0

dependencies: added serde and serde_json

60facae

feat: added first test flow to suite

470186c

feat: made tracer print optional

b616f78

feat: exported RuntimeErrors name and message

ed42674

feat: first correct flow test cases

720baf9

feat: added readme

8ad8b83

feat: added comparison of results to test suite

a7bfd92

ref: cargo fix

7d2c92e

ref: cargo clippy

42351da

ref: cargo fmt

6701195

feat: reworked execution logic

3619a47

feat: new reference logic impl

a8f1187

feat: added tracing

0783904

feat: wip new implementation of array functions

d1b7d42

dependencies: updated to tucana 0.0.53

72d5cb4

raphael-goetz added 17 commits March 14, 2026 20:57

feat: added wip flow test execution suite

a390f51

dependencies: added serde and serde_json

0fcd17f

feat: added first test flow to suite

2f15a41

feat: made tracer print optional

7f035c4

feat: exported RuntimeErrors name and message

08c2459

feat: first correct flow test cases

9ef8f83

feat: added readme

3c66e32

feat: added comparison of results to test suite

eb815b5

ref: cargo fix

92bc170

ref: cargo clippy

43d27ef

ref: cargo fmt

2264a44

Merge branch 'major-rework' of https://github.com/code0-tech/taurus i…

5fbaab0

…nto major-rework

feat: added more detail options for tracer

7b07b0c

feat: added for each test flow

fe61eab

fix: renamed core crate to taurus core to fix naming errors because c…

6db39d1

…rate was referncing core lib

feat: added mock functions for iterator that accepts some sort of fun…

73ea8c3

…ctions

feat: added test suite to github workflow

9311a12

raphael-goetz requested a review from Copilot March 16, 2026 14:23

Copilot started reviewing on behalf of raphael-goetz March 16, 2026 14:23 View session

feat: cargo clippy

b572593

This was linked to issues Mar 16, 2026

Propper Debuging Mode #98

Closed

Update Handling of Reference Values #113

Closed

std::number::math Panics when Second Number is Bigger then First #117

Closed

Copilot AI reviewed Mar 16, 2026

View reviewed changes

raphael-goetz and others added 2 commits March 16, 2026 15:36

Potential fix for pull request finding

511144d

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> Signed-off-by: Raphael Götz <52959657+raphael-goetz@users.noreply.github.com>

Potential fix for pull request finding

60b0465

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> Signed-off-by: Raphael Götz <52959657+raphael-goetz@users.noreply.github.com>

raphael-goetz requested a review from Copilot March 16, 2026 14:38

Copilot started reviewing on behalf of raphael-goetz March 16, 2026 14:39 View session

Copilot AI reviewed Mar 16, 2026

View reviewed changes

raphael-goetz merged commit d1fbeb1 into main Mar 16, 2026
5 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New Reference Logic #121

New Reference Logic #121
raphael-goetz merged 59 commits intomainfrom
major-rework

raphael-goetz commented Mar 16, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

raphael-goetz commented Mar 16, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants