HORIZON HASKELLDocslts/ghc-9.10.xc74966e2026-09-27Search names, modules, packages, or :: a typeCtrl K

GHC 9.10.3 · lts/ghc-9.10.x · c74966e · 2026-09-27

Modulebase-4.20.2.0Haskell2010

Debug.Trace

Functions for tracing and monitoring execution.

These can be useful for investigating bugs or performance problems. They should not be used in production code.

  • 17 values
  • Packagebase-4.20.2.0
  • Exports17
  • LanguageHaskell2010
  • LicenceBSD-3-Clause
  • SourceTrace.hs

Tracing

11 declarations

The trace, traceShow and traceIO functions print messages to an output stream. They are intended for "printf debugging", that is: tracing the flow of execution and printing interesting values.

All these functions evaluate the message completely before printing it; so if the message is not fully defined, none of it will be printed.

The usual output stream is GHC.Internal.System.IO.stderr. For Windows GUI applications (that have no stderr) the output is directed to the Windows debug console. Some implementations of these functions may decorate the string that's output to indicate that you're tracing.

valuetrace :: String -> a -> a
#

The trace function outputs the trace message given as its first argument, before returning the second argument as its result.

For example, this returns the value of f x and outputs the message to stderr. Depending on your terminal (settings), they may or may not be mixed.

Example2 expressions
let x = 123; f = showtrace ("calling f with x = " ++ show x) (f x)calling f with x = 123"123"

The trace function should only be used for debugging, or for monitoring execution. The function is not referentially transparent: its type indicates that it is a pure function but it has the side effect of outputting the trace message.

valuetraceId :: String -> String
#

Like trace but returns the message instead of a third value.

Example1 expression
traceId "hello"hello"hello"
valuetraceShow :: Show a => a -> b -> b
#

Like trace, but uses show on the argument to convert it to a String.

This makes it convenient for printing the values of interesting variables or expressions inside a function. For example, here we print the values of the variables x and y:

Example1 expression
let f x y = traceShow ("x", x, "y", y) (x + y) in f (1+2) 5("x",3,"y",5)8

Note in this example we also create simple labels just by including some strings.

valuetraceShowId :: Show a => a -> a
#

Like traceShow but returns the shown value instead of a third value.

Example1 expression
traceShowId (1+2+3, "hello" ++ "world")(6,"helloworld")(6,"helloworld")
valuetraceWith :: (a -> String) -> a -> a
#

Like trace, but outputs the result of calling a function on the argument.

Example1 expression
traceWith fst ("hello","world")hello("hello","world")
valuetraceShowWith :: Show b => (a -> b) -> a -> a
#

Like traceWith, but uses show on the result of the function to convert it to a String.

Example1 expression
traceShowWith length [1,2,3]3[1,2,3]
valuetraceStack :: String -> a -> a
#

like trace, but additionally prints a call stack if one is available.

In the current GHC implementation, the call stack is only available if the program was compiled with -prof; otherwise traceStack behaves exactly like trace. Entries in the call stack correspond to SCC annotations, so it is a good idea to use -fprof-auto or -fprof-auto-calls to add SCC annotations automatically.

valuetraceIO :: String -> IO ()
#

The traceIO function outputs the trace message from the IO monad. This sequences the output with respect to other IO actions.

valuetraceM :: Applicative f => String -> f ()
#

Like trace but returning unit in an arbitrary Applicative context. Allows for convenient use in do-notation.

Note that the application of traceM is not an action in the Applicative context, as traceIO is in the IO type. While the fresh bindings in the following example will force the traceM expressions to be reduced every time the do-block is executed, traceM "not crashed" would only be reduced once, and the message would only be printed once. If your monad is in MonadIO, liftIO . traceIO may be a better option.

Example1 expression
:{do    x <- Just 3    traceM ("x: " ++ show x)    y <- pure 12    traceM ("y: " ++ show y)    pure (x*2 + y):}x: 3y: 12Just 18
valuetraceShowM :: (Show a, Applicative f) => a -> f ()
#

Like traceM, but uses show on the argument to convert it to a String.

Example1 expression
:{do    x <- Just 3    traceShowM x    y <- pure 12    traceShowM y    pure (x*2 + y):}312Just 18

Eventlog tracing

4 declarations

Eventlog tracing is a performance profiling system. These functions emit extra events into the eventlog. In combination with eventlog profiling tools these functions can be used for monitoring execution and investigating performance problems.

Currently only GHC provides eventlog profiling, see the GHC user guide for details on how to use it. These function exists for other Haskell implementations but no events are emitted. Note that the string message is always evaluated, whether or not profiling is available or enabled.

valuetraceEvent :: String -> a -> a
#

The traceEvent function behaves like trace with the difference that the message is emitted to the eventlog, if eventlog profiling is available and enabled at runtime.

It is suitable for use in pure code. In an IO context use traceEventIO instead.

Note that when using GHC's SMP runtime, it is possible (but rare) to get duplicate events emitted if two CPUs simultaneously evaluate the same thunk that uses traceEvent.

Execution phase markers

2 declarations

When looking at a profile for the execution of a program we often want to be able to mark certain points or phases in the execution and see that visually in the profile.

For example, a program might have several distinct phases with different performance or resource behaviour in each phase. To properly interpret the profile graph we really want to see when each phase starts and ends.

Markers let us do this: we can annotate the program to emit a marker at an appropriate point during execution and then see that in a profile.

Currently this feature is only supported in GHC by the eventlog tracing system, but in future it may also be supported by the heap profiling or other profiling tools. These function exists for other Haskell implementations but they have no effect. Note that the string message is always evaluated, whether or not profiling is available or enabled.

valuetraceMarker :: String -> a -> a
#

The traceMarker function emits a marker to the eventlog, if eventlog profiling is available and enabled at runtime. The String is the name of the marker. The name is just used in the profiling tools to help you keep clear which marker is which.

This function is suitable for use in pure code. In an IO context use traceMarkerIO instead.

Note that when using GHC's SMP runtime, it is possible (but rare) to get duplicate events emitted if two CPUs simultaneously evaluate the same thunk that uses traceMarker.