Add a way to generate callstack trace message via plugin side generation #7148

SeungheonOh · 2025-06-16T10:32:25Z

closes #7146. Add plugin option to generate callstack trace message. Currently this won't handle cases when a function has PlutusTx.Trace.callStack and that function is called in multiple places. It will just use callstack of the first occurance :/ (working on it. more detail on first commit message).

I redid everything, now it will have correct callstack, but with some execution cost overhead. I'm not too unhappy about this since callstack in general is almost exclusively used for debug scripts anyways.

Example

-- SillyCalss.hs
class SillyClass a where
  sillyThing :: a -> Integer

instance SillyClass Integer where
  sillyThing = traceError callStack

-- Demo.hs
rob :: Integer -> Integer -> Integer
rob x = sillyThing @Integer

-- entrypoint
tom :: Integer -> Integer
tom = rob (x PTx.+ 20)

tom:demo/Demo.hs:44:1-44:3
\-rob:demo/Demo.hs:32:1-32:3
\-$fSillyClassInteger
\-$fSillyClassInteger_$csillyThing

Added `generate-callstack` option, which, when turned on, will replace every occurances of `PlutusTx.Trace.callStack` with the callstack message at its relative position. This is "rudimentary" because it won't correctly handled when `PlutusTx.Trace.callStack` is called inside of function, call it `bob`, which gets called in multiple locations. It will use callstack when `bob` was called the first time and use that exact callstack in any subsequent calls of `bob` which can be very misleading. This, however, is the best we can get without needing to pass around callstack values at UPLC level or generating multiple copies of `bob` for all of its unique relative positions.

github-actions · 2025-06-16T10:38:09Z

PR Preview Action v1.6.1
🚀 View preview at https://IntersectMBO.github.io/plutus/pr-preview/pr-7148/
Built to branch `gh-pages` at 2025-06-17 20:25 UTC. Preview will be ready when the GitHub Pages deployment is complete.

SeungheonOh · 2025-06-16T19:38:32Z

Figured out a way to do "real" callstack, with some runtime overhead. This will handle something like this correctly:

silly, silly2, silly3, silly4 :: Integer -> Integer
silly x = silly2 x
silly2 x = silly3 x
silly3 x = silly4 x
silly4 x = bob

bob, rob, cob :: Integer
bob = traceError callStack
rob = traceError callStack
cob = traceError callStack

tom :: Integer -> Integer
tom 1 = silly 1
tom 2 = rob
tom 3 = cob
tom _ = 42

[||tom 1||]
tom:demo/Demo.hs:62:1-62:3
\-silly:demo/Demo.hs:51:1-51:5
\-silly2:demo/Demo.hs:52:1-52:6
\-silly3:demo/Demo.hs:53:1-53:6
\-silly4:demo/Demo.hs:54:1-54:6
\-bob:demo/Demo.hs:57:1-57:3

[||tom 2||]
tom:demo/Demo.hs:62:1-62:3
\-rob:demo/Demo.hs:58:1-58:3

[||tom 3||]
tom:demo/Demo.hs:62:1-62:3
\-cob:demo/Demo.hs:59:1-59:3

SeungheonOh · 2025-06-17T18:14:29Z

plutus-tx-plugin/src/PlutusTx/Compiler/Expr.hs

+            -- FIXME: Variables will miss span information when the module they are defined
+            -- in is loaded from cache instead of getting compiled.


Is there a workaround for this?

SeungheonOh · 2025-06-17T18:51:21Z

plutus-tx-plugin/test/Plugin/CallStack/9.6/func01.eval.golden

+\-nestedLinear:test/Plugin/CallStack/Spec.hs:63:1-63:12
+\-nestedLinear2:test/Plugin/CallStack/Spec.hs:64:1-64:13
+\-nestedLinear3:test/Plugin/CallStack/Spec.hs:65:1-65:13
+\-nestedLinear4:test/Plugin/CallStack/Spec.hs:66:1-66:13


The span provided points to the definition of executed function instead of location of invocation. Is this fine?

... I think it's fine

This PR is meant to achieve the goal stated in the issue it is associated with. From the issue description "is useful for determining which part of validator in particular cause the failure." I deduce that goal is for on-chain program authors to be able to determine "which part of a validator causes the failure".

Given two possible implementations:

One that prints locations of invocations.

Another that prints locations of definitions.

I'd say that its straightforward how the former one could help authors to determine the failing validator part, and is not straightforward if the latter always helps.

Is there a fundamental reason why we can't have invocation locations?
Please document it in the note, if that's the case.

Unisay · 2025-06-19T08:00:34Z

plutus-tx-plugin/changelog.d/20250618_025657_seungheon.ooh_7146.md

@@ -0,0 +1,4 @@
+### Added
+- `generate-callstack` option: When enabled, the plugin replaces every occurrence of `PlutusTx.Trace.callStack` with a `BuiltinString` containing the callstack at that relative point. Enabling this option incurs execution-cost overhead: it adds an extra callstack argument to all user-defined functions, and every time they are invoked it requires having extra application node for providing the current callstack. When `gennerate-callstack` is not enabled, `PlutusTx.Trace.callStack` is string "<CallStack>".


Suggested change

- `generate-callstack` option: When enabled, the plugin replaces every occurrence of `PlutusTx.Trace.callStack` with a `BuiltinString` containing the callstack at that relative point. Enabling this option incurs execution-cost overhead: it adds an extra callstack argument to all user-defined functions, and every time they are invoked it requires having extra application node for providing the current callstack. When `gennerate-callstack` is not enabled, `PlutusTx.Trace.callStack` is string "<CallStack>".

- `generate-callstack` option: When enabled, the plugin replaces every occurrence of `PlutusTx.Trace.callStack` with a `BuiltinString` containing the callstack at that relative point. Enabling this option incurs execution-cost overhead: it adds an extra callstack argument to all user-defined functions, and every time they are invoked it requires having extra application node for providing the current callstack. When `generate-callstack` is not enabled, `PlutusTx.Trace.callStack` is string "<CallStack>".

ana-pantilie

Nice! LGTM, with the caveat that I'm not very familiar with the code in Expr. Maybe somebody else should approve as well before you merge.

Unisay · 2025-06-23T12:49:32Z

plutus-tx-plugin/src/PlutusTx/Compiler/Types.hs

  }

+lastCallStackName :: MonadState CompileState m => m (Maybe (PIR.Name))


Suggested change

lastCallStackName :: MonadState CompileState m => m (Maybe (PIR.Name))

lastCallStackName :: MonadState CompileState m => m (Maybe PIR.Name)

Unisay · 2025-06-23T12:54:27Z

plutus-tx-plugin/src/PlutusTx/Options.hs

@@ -309,6 +310,9 @@ pluginOptions =
            rest <- many (alphaNumChar <|> char '_' <|> char '\\')
            pure (firstC : rest)
       in (k, PluginOption typeRep (plcParserOption p k) posCertify desc [])
+    , let k = "generate-callstack"
+          desc = "Generate callstack string and replace 'callStack' with callstack at the location."


What does the "replace 'callStack' with callstack at the location." part mean for a compiler user?
I don't understand it myself, don't think that end-users will get it either.

Unisay · 2025-06-23T13:08:36Z

plutus-tx-plugin/test/Plugin/CallStack/9.6/func02.eval.golden

+\-nestedLinear2:test/Plugin/CallStack/Spec.hs:64:1-64:13
+\-nestedLinear3:test/Plugin/CallStack/Spec.hs:65:1-65:13
+\-nestedLinear4:test/Plugin/CallStack/Spec.hs:66:1-66:13


Why did you use \ backslash? I'd find it more useful to have some notion of a direction here: should I read it top to bottom or vice-versa? Maybe 1. 2. 3. is better then?

SeungheonOh · 2025-07-02T06:03:32Z

superseded by #7178 unless someone actually prefers this method

SeungheonOh temporarily deployed to github-pages June 16, 2025 10:32 — with GitHub Actions Inactive

WIP try "actual" tracing that require runtime overhead

cba8096

SeungheonOh temporarily deployed to github-pages June 16, 2025 19:29 — with GitHub Actions Inactive

SeungheonOh requested a review from Unisay June 17, 2025 13:16

SeungheonOh self-assigned this Jun 17, 2025

SeungheonOh added Plinth User experience Plinth Plugin labels Jun 17, 2025

Remove old stactic callstack, add runtime base callstack

d6258a1

SeungheonOh temporarily deployed to github-pages June 17, 2025 18:11 — with GitHub Actions Inactive

SeungheonOh requested a review from ana-pantilie June 17, 2025 18:11

SeungheonOh force-pushed the sho/7146 branch from 57d4ebc to 032b203 Compare June 17, 2025 18:19

SeungheonOh temporarily deployed to github-pages June 17, 2025 18:19 — with GitHub Actions Inactive

SeungheonOh commented Jun 17, 2025

View reviewed changes

SeungheonOh force-pushed the sho/7146 branch from 032b203 to 461fa9e Compare June 17, 2025 18:29

SeungheonOh temporarily deployed to github-pages June 17, 2025 18:29 — with GitHub Actions Inactive

add changelog

d3a9594

SeungheonOh force-pushed the sho/7146 branch from 461fa9e to d3a9594 Compare June 17, 2025 18:35

SeungheonOh temporarily deployed to github-pages June 17, 2025 18:35 — with GitHub Actions Inactive

SeungheonOh commented Jun 17, 2025

View reviewed changes

SeungheonOh mentioned this pull request Jun 18, 2025

Incorrect callstack trace when function is used multiple times in different branches Plutonomicon/plutarch-plutus#838

Open

Unisay reviewed Jun 19, 2025

View reviewed changes

ana-pantilie approved these changes Jun 23, 2025

View reviewed changes

Unisay reviewed Jun 23, 2025

View reviewed changes

SeungheonOh changed the title ~~Add a way to generate callstack trace message~~ Add a way to generate callstack trace message via plugin side generation Jul 1, 2025

SeungheonOh mentioned this pull request Jul 2, 2025

Call trace for failing evaluation via new emitter mode #7178

Open

SeungheonOh closed this Jul 2, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add a way to generate callstack trace message via plugin side generation #7148

Add a way to generate callstack trace message via plugin side generation #7148

Uh oh!

SeungheonOh commented Jun 16, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jun 16, 2025 •

edited

Loading

Built to branch `gh-pages` at 2025-06-17 20:25 UTC.
Preview will be ready when the GitHub Pages deployment is complete.

Uh oh!

SeungheonOh commented Jun 16, 2025

Uh oh!

SeungheonOh Jun 17, 2025

Uh oh!

SeungheonOh Jun 17, 2025 •

edited

Loading

Uh oh!

Unisay Jun 23, 2025

Uh oh!

Unisay Jun 19, 2025

Uh oh!

ana-pantilie left a comment

Uh oh!

Unisay Jun 23, 2025

Uh oh!

Unisay Jun 23, 2025

Uh oh!

Unisay Jun 23, 2025

Uh oh!

SeungheonOh commented Jul 2, 2025

Uh oh!

Uh oh!

		-- FIXME: Variables will miss span information when the module they are defined
		-- in is loaded from cache instead of getting compiled.

		@@ -0,0 +1,4 @@
		### Added
		- `generate-callstack` option: When enabled, the plugin replaces every occurrence of `PlutusTx.Trace.callStack` with a `BuiltinString` containing the callstack at that relative point. Enabling this option incurs execution-cost overhead: it adds an extra callstack argument to all user-defined functions, and every time they are invoked it requires having extra application node for providing the current callstack. When `gennerate-callstack` is not enabled, `PlutusTx.Trace.callStack` is string "<CallStack>".

		}

		lastCallStackName :: MonadState CompileState m => m (Maybe (PIR.Name))

Add a way to generate callstack trace message via plugin side generation #7148

Add a way to generate callstack trace message via plugin side generation #7148

Uh oh!

Conversation

SeungheonOh commented Jun 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Jun 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Built to branch gh-pages at 2025-06-17 20:25 UTC. Preview will be ready when the GitHub Pages deployment is complete.

Uh oh!

SeungheonOh commented Jun 16, 2025

Uh oh!

SeungheonOh Jun 17, 2025

Choose a reason for hiding this comment

Uh oh!

SeungheonOh Jun 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Unisay Jun 23, 2025

Choose a reason for hiding this comment

Uh oh!

Unisay Jun 19, 2025

Choose a reason for hiding this comment

Uh oh!

ana-pantilie left a comment

Choose a reason for hiding this comment

Uh oh!

Unisay Jun 23, 2025

Choose a reason for hiding this comment

Uh oh!

Unisay Jun 23, 2025

Choose a reason for hiding this comment

Uh oh!

Unisay Jun 23, 2025

Choose a reason for hiding this comment

Uh oh!

SeungheonOh commented Jul 2, 2025

Uh oh!

Uh oh!

SeungheonOh commented Jun 16, 2025 •

edited

Loading

github-actions bot commented Jun 16, 2025 •

edited

Loading

Built to branch `gh-pages` at 2025-06-17 20:25 UTC.
Preview will be ready when the GitHub Pages deployment is complete.

SeungheonOh Jun 17, 2025 •

edited

Loading