:orphan:
# PetscError
Routine that is called when an error has been detected, usually called through the macro SETERRQ(PETSC_COMM_SELF,). 
## Synopsis
```
#include "petscsys.h" 
PetscErrorCode PetscError(MPI_Comm comm, int line, const char *func, const char *file, PetscErrorCode n, PetscErrorType p, const char *mess, ...)
```
Collective


## Input Parameters

- ***comm -*** communicator over which error occurred.  ALL ranks of this communicator MUST call this routine
- ***line -*** the line number of the error (indicated by __LINE__)
- ***func -*** the function name in which the error was detected
- ***file -*** the file in which the error was detected (indicated by __FILE__)
- ***n -*** the generic error number
- ***p -*** `PETSC_ERROR_INITIAL` indicates the error was initially detected, `PETSC_ERROR_REPEAT` indicates this is a traceback from a previously detected error
- ***mess -*** formatted message string - aka printf



## Options Database Keys

- ***-error_output_stdout -*** output the error messages to stdout instead of the default stderr
- ***-error_output_none -*** do not output the error messages





## Notes
PETSc error handling is done with error return codes. A non-zero return indicates an error
was detected. The return-value of this routine is what is ultimately returned by
`SETERRQ()`.

Note that numerical errors (potential divide by zero, for example) are not managed by the
error return codes; they are managed via, for example, `KSPGetConvergedReason()` that
indicates if the solve was successful or not. The option `-ksp_error_if_not_converged`, for
example, turns numerical failures into hard errors managed via `PetscError()`.

PETSc provides a rich supply of error handlers, see the list below, and users can also
provide their own error handlers.

If the user sets their own error handler (via `PetscPushErrorHandler()`) they may return any
arbitrary value from it, but are encouraged to return nonzero values. If the return value is
zero, `SETERRQ()` will ignore the value and return `PETSC_ERR_RETURN` (a nonzero value)
instead.

Most users need not directly use this routine and the error handlers, but can instead use
the simplified interface `PetscCall()` or `SETERRQ()`.


## Fortran Note
This routine is used differently from Fortran
```none
PetscError(MPI_Comm comm, PetscErrorCode n, PetscErrorType p, char *message)
```


## Developer Note
Since this is called after an error condition it should not be calling any error handlers (currently it ignores any error codes)
BUT this routine does call regular PETSc functions that may call error handlers, this is problematic and could be fixed by never calling other PETSc routines
but this annoying.


## See Also
 `PetscErrorCode`, `PetscPushErrorHandler()`, `PetscPopErrorHandler()`, `PetscTraceBackErrorHandler()`, `PetscAbortErrorHandler()`, `PetscMPIAbortErrorHandler()`,
`PetscReturnErrorHandler()`, `PetscAttachDebuggerErrorHandler()`, `PetscEmacsClientErrorHandler()`,
`SETERRQ()`, `PetscCall()`, `CHKMEMQ`, `SETERRQ()`, `SETERRQ()`, `PetscErrorMessage()`, `PETSCABORT()`

## Level
intermediate

## Location
<A HREF="PETSC_DOC_OUT_ROOT_PLACEHOLDER/src/sys/error/err.c.html#PetscError">src/sys/error/err.c</A>

## Examples
<A HREF="PETSC_DOC_OUT_ROOT_PLACEHOLDER/src/vec/vec/tutorials/ex21.c.html">src/vec/vec/tutorials/ex21.c</A><BR>


---
[Edit on GitLab](https://gitlab.com/petsc/petsc/-/edit/release/src/sys/error/err.c)


[Index of all Sys routines](index.md)  
[Table of Contents for all manual pages](/manualpages/index.md)  
[Index of all manual pages](/manualpages/singleindex.md)