Bug: Cannot construct a value of type: PlutusTx.Builtins.Internal.BuiltinUnit
#7162
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,135 @@ | ||
| # Investigation: Disabling mkSimplPass in PlutusTx Plugin | ||
|
|
||
| ## Context | ||
|
|
||
| This investigation was conducted to understand whether disabling the `mkSimplPass` in the PlutusTx plugin would improve error messages for stage violations, specifically the confusing "Cannot construct a value of type: BuiltinUnit" error reported in [plutus-private#1626](https://github.com/IntersectMBO/plutus-private/issues/1626). | ||
|
|
||
| ## Background | ||
|
|
||
| The `mkSimplPass` was originally added to work around [GHC #16615](https://gitlab.haskell.org/ghc/ghc/-/issues/16615), where local bindings lack unfoldings when Core plugins run early in the compilation pipeline. The simplifier pass with `sm_pre_inline = True` generates these unfoldings by inlining bindings that occur exactly once unconditionally. | ||
|
|
||
| ### Why mkSimplPass Was Needed | ||
|
|
||
| From Note [GHC.sm_pre_inline] in the code: | ||
| - The plugin requires certain functions to be fully applied | ||
| - Example: it can handle `noinline @(String -> BuiltinString) stringToBuiltinString "a"` but not `let f = noinline @(String -> BuiltinString) stringToBuiltinString in f "a"` | ||
| - Pre-inlining solves this by inlining single-use bindings before the plugin runs | ||
|
|
||
| ## Investigation Method | ||
|
|
||
| 1. Disabled `mkSimplPass` in `PlutusTx.Plugin.install` | ||
| 2. Attempted to build the plutus-tx-plugin test suite | ||
| 3. Categorized and analyzed all compilation failures | ||
| 4. Assessed whether failures could be fixed without the simplifier pass | ||
|
|
||
| ## Key Findings | ||
|
|
||
| ### ✅ Primary Goal Achieved: Better Error Messages | ||
|
|
||
| **Before (with mkSimplPass):** | ||
| ``` | ||
| Error: Unsupported feature: Cannot construct a value of type: BuiltinUnit | ||
| Context: Compiling expr: BuiltinUnit | ||
| Context: Compiling expr: BuiltinUnit ipv | ||
| ``` | ||
|
|
||
| **After (without mkSimplPass):** | ||
| ``` | ||
| Error: Reference to a name which is not a local, a builtin, or an external INLINABLE function: Variable validator1 | ||
| No unfolding | ||
| Context: Compiling expr: validator1 | ||
| Context: Compiling expr at: test/BuiltinUnit/Spec.hs:74:11-36 | ||
| ``` | ||
|
|
||
| The new error message clearly indicates the stage violation problem and points to the exact location, making it much easier for users to understand and fix. | ||
|
|
||
| ### Failure Categories | ||
|
|
||
| Without `mkSimplPass`, we observed three main categories of failures: | ||
|
|
||
| #### 1. Stage Violations (✅ Fixed) | ||
| - **BuiltinUnit test cases**: Now produce clear "No unfolding" errors instead of confusing builtin constructor errors | ||
| - **This is the desired behavior** - code that violates PlutusTx compilation rules should fail with meaningful messages | ||
|
|
||
| #### 2. Missing Unfoldings for INLINEABLE Functions (🔧 Potentially Fixable) | ||
| ``` | ||
| Error: Reference to a name which is not a local, a builtin, or an external INLINABLE function: Variable Plugin.Coverage.Spec.fun | ||
| OtherCon [] | ||
| ``` | ||
| - **Root cause**: GHC #16615 - local bindings lack unfoldings without simplifier | ||
| - **Pattern**: Functions marked `{-# INLINEABLE #-}` but defined after their use site | ||
| - **Solutions**: | ||
| - Move functions to top-level | ||
| - Generate minimal unfoldings during desugaring (GHC enhancement needed) | ||
| - Create lightweight unfolding pass (lighter than full simplifier) | ||
|
There was a problem hiding this comment. is this possible without making changes to GHC itself? From what I can see(which can be very wrong), we can only really change the plugin pass and the plugin pass doesn't have access to other stages of compilation. This would be very easy if we have whole compilation pipeline under control. |
||
|
|
||
| #### 3. String Literal Processing Issues (🔧 Potentially Fixable) | ||
| ``` | ||
| Error: Unsupported feature: Literal string (maybe you need to use OverloadedStrings) | ||
| Context: Compiling expr: "f0d1"# | ||
| ``` | ||
| - **Root cause**: String literals not being processed properly without pre-inlining | ||
| - **Pattern**: Hex string literals in ByteString contexts | ||
| - **Solutions**: | ||
| - Pre-process string literals before plugin runs | ||
| - Better handling of `OverloadedStrings` in plugin | ||
| - Add specific support for hex string literal patterns | ||
|
|
||
| ## Feasibility Assessment | ||
|
|
||
| ### ✅ **Highly Feasible** | ||
|
|
||
| The investigation shows that disabling `mkSimplPass` is **feasible** for several reasons: | ||
|
|
||
| 1. **Primary goal achieved**: Stage violations now produce clear, actionable error messages | ||
| 2. **Most failures are fixable**: The breaking changes are due to missing unfoldings, not fundamental incompatibilities | ||
| 3. **Targeted solutions possible**: Each failure category has potential solutions that don't require the full complexity of the current simplifier pass | ||
| 4. **Performance benefits**: Removing the simplifier pass would improve compilation performance | ||
|
|
||
| ### Recommended Approach | ||
|
|
||
| 1. **Phase 1**: Implement targeted fixes for the most common failure patterns: | ||
| - Add lightweight unfolding generation for `INLINEABLE` functions | ||
| - Improve string literal handling | ||
| - Update problematic test code organization | ||
|
|
||
| 2. **Phase 2**: Gradually remove `mkSimplPass` once core issues are resolved: | ||
| - Start with a flag to disable the pass for testing | ||
| - Monitor performance and correctness | ||
| - Full removal once confidence is high | ||
|
|
||
| 3. **Phase 3**: Enhance error messages further: | ||
| - Add specific detection and messaging for stage violations | ||
| - Provide suggestions for fixing common problems | ||
|
|
||
| ## Impact Analysis | ||
|
|
||
| ### Benefits | ||
| - **Much clearer error messages** for stage violations | ||
| - **Better compilation performance** (no unnecessary simplification) | ||
| - **Reduced complexity** in the plugin pipeline | ||
| - **Better alignment** with PlutusTx compilation model | ||
|
|
||
| ### Risks | ||
| - **Some legitimate code patterns may break** initially | ||
| - **Requires careful migration** of existing code | ||
| - **May need GHC enhancements** for optimal solution (addressing GHC #16615) | ||
|
|
||
| ## Conclusion | ||
|
|
||
| Disabling `mkSimplPass` is a **promising approach** that achieves the primary goal of improving error messages while being technically feasible. The investigation demonstrates that most failures can be addressed through targeted solutions rather than requiring the full simplifier pass. | ||
|
|
||
| The trade-off between better error messages and some initial compatibility issues is favorable, especially given that the breaking patterns often represent code that violates PlutusTx compilation rules anyway. | ||
|
|
||
| ## Test Results Summary | ||
|
|
||
| - **BuiltinUnit.Spec**: ✅ Compiles successfully, produces clear error messages | ||
| - **Plugin tests with unfolding issues**: ❌ Need targeted fixes | ||
| - **String literal tests**: ❌ Need preprocessing improvements | ||
| - **Overall plugin functionality**: 🔧 Maintainable with targeted improvements | ||
|
|
||
| --- | ||
|
|
||
| *Investigation conducted on: 2025-01-25* | ||
| *Branch: yura/Cannot-construct-a-value-of-type* | ||
| *Related Issues: [plutus-private#1626](https://github.com/IntersectMBO/plutus-private/issues/1626), [GHC #16615](https://gitlab.haskell.org/ghc/ghc/-/issues/16615)* | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,106 @@ | ||
| {-# LANGUAGE BlockArguments #-} | ||
| {-# LANGUAGE DataKinds #-} | ||
| {-# LANGUAGE NoImplicitPrelude #-} | ||
| {-# LANGUAGE Strict #-} | ||
| {-# LANGUAGE TemplateHaskell #-} | ||
| {-# LANGUAGE TypeApplications #-} | ||
| {-# OPTIONS_GHC -ddump-simpl-iterations -dsuppress-all #-} | ||
| {-# OPTIONS_GHC -fno-float-in #-} | ||
| {-# OPTIONS_GHC -fno-full-laziness #-} | ||
| {-# OPTIONS_GHC -fno-ignore-interface-pragmas #-} | ||
| {-# OPTIONS_GHC -fno-local-float-out #-} | ||
| {-# OPTIONS_GHC -fno-omit-interface-pragmas #-} | ||
| {-# OPTIONS_GHC -fno-spec-constr #-} | ||
| {-# OPTIONS_GHC -fno-specialise #-} | ||
| {-# OPTIONS_GHC -fno-strictness #-} | ||
| {-# OPTIONS_GHC -fno-unbox-small-strict-fields #-} | ||
| {-# OPTIONS_GHC -fno-unbox-strict-fields #-} | ||
| {-# OPTIONS_GHC -fplugin PlutusTx.Plugin #-} | ||
|
|
||
| module BuiltinUnit.Spec where | ||
|
|
||
| import PlutusTx.Prelude | ||
| import Prelude (IO, seq) | ||
|
|
||
| import Control.Lens (view) | ||
| import PlutusTx (CompiledCode, compile, getPlcNoAnn) | ||
| import PlutusTx.Builtins.Internal (unitval) | ||
| import Test.Tasty (TestTree, testGroup) | ||
| import Test.Tasty.HUnit (Assertion, testCase) | ||
| import UntypedPlutusCore (progTerm) | ||
|
|
||
| tests :: TestTree | ||
| tests = | ||
| testGroup | ||
| "BuiltinUnit" | ||
| [ testCase "error ()" do assertCompiledTerm code1 | ||
| , testCase "unitval" do assertCompiledTerm code2 | ||
| , testCase "locally defined constructor" do assertCompiledTerm code3 | ||
| , testCase "toOpaque ()" do assertCompiledTerm code4 | ||
| ] | ||
|
|
||
| {- GHC Core after simplification: | ||
|
|
||
| code1 = case error () of | ||
| validator1_X0 { BuiltinUnit ipv_smFH -> plc Proxy validator1_X0 } | ||
|
|
||
| code2 = case unitval of | ||
| validator2_X0 { BuiltinUnit ipv_smFJ -> plc Proxy validator2_X0 } | ||
|
|
||
| code3 = case unitval of | ||
| builtinUnit_X0 { BuiltinUnit ipv_smFL -> plc Proxy builtinUnit_X0 } | ||
|
|
||
| code4 = case toOpaque $fHasToOpaque()BuiltinUnit () of | ||
| validator4_X0 { BuiltinUnit ipv_smGp -> plc Proxy validator4_X0 } | ||
|
|
||
| Compilation error: | ||
|
|
||
| <no location info>: error: | ||
| GHC Core to PLC plugin: | ||
| Error: Unsupported feature: | ||
| Cannot construct a value of type: BuiltinUnit | ||
| Note: GHC can generate these unexpectedly, you may need | ||
| '-fno-strictness', '-fno-specialise', '-fno-spec-constr', | ||
| '-fno-unbox-strict-fields', or '-fno-unbox-small-strict-fields'. | ||
| Context: Compiling expr: BuiltinUnit | ||
| Context: Compiling expr: BuiltinUnit ipv | ||
| Context: Compiling definition of: validator1 | ||
| Context: Compiling expr: validator1 | ||
| Context: Compiling expr at: test/BuiltinUnit/Spec.hs:75:11-36 | ||
| Context: Compiling expr: validator1 | ||
| -} | ||
|
|
||
| code1 :: CompiledCode BuiltinUnit | ||
| code1 = $$(compile [||validator1||]) | ||
| where | ||
| validator1 :: BuiltinUnit | ||
| validator1 = PlutusTx.Prelude.error () | ||
| {-# INLINEABLE validator1 #-} | ||
|
|
||
| code2 :: CompiledCode BuiltinUnit | ||
| code2 = $$(compile [||validator2||]) | ||
| where | ||
| validator2 :: BuiltinUnit | ||
| validator2 = unitval | ||
| {-# INLINEABLE validator2 #-} | ||
|
|
||
| code3 :: CompiledCode BuiltinUnit | ||
| code3 = $$(compile [||validator3||]) | ||
| where | ||
| validator3 :: BuiltinUnit | ||
| validator3 = builtinUnit | ||
| {-# INLINEABLE validator3 #-} | ||
|
|
||
| builtinUnit :: BuiltinUnit | ||
| builtinUnit = unitval | ||
| {-# INLINEABLE builtinUnit #-} | ||
|
|
||
| code4 :: CompiledCode BuiltinUnit | ||
| code4 = $$(compile [||validator4||]) | ||
| where | ||
| validator4 :: BuiltinUnit | ||
| validator4 = toOpaque () | ||
| {-# INLINEABLE validator4 #-} | ||
|
|
||
| assertCompiledTerm :: CompiledCode a -> Assertion | ||
| assertCompiledTerm code = view progTerm (getPlcNoAnn code) `seq` return @IO () |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's probably also the case for function marked inlineable but defined on the other module.