These procedures are used to manipulate two Tcl global variables
that hold information about errors.
The variable errorInfo holds a stack trace of the
operations that were in progress when an error occurred,
and is intended to be human-readable.
The variable errorCode holds a list of items that
are intended to be machine-readable.
The first item in errorCode identifies the class of
error that occurred
(e.g. POSIX means an error occurred in a POSIX system call)
and additional elements in errorCode hold additional pieces
of information that depend on the class.
See the Tcl overview manual entry for details on the various
formats for errorCode.
The errorInfo variable is gradually built up as an
error unwinds through the nested operations.
Each time an error code is returned to Tcl_EvalObj
(or Tcl_Eval, which calls Tcl_EvalObj)
it calls the procedure Tcl_AddObjErrorInfo to add
additional text to errorInfo describing the
command that was being executed when the error occurred.
By the time the error has been passed all the way back
to the application, it will contain a complete trace
of the activity in progress when the error occurred.
It is sometimes useful to add additional information to
errorInfo beyond what can be supplied automatically
by Tcl_EvalObj.
Tcl_AddObjErrorInfo may be used for this purpose:
its message and length arguments describe an additional
string to be appended to errorInfo.
For example, the source command calls Tcl_AddObjErrorInfo
to record the name of the file being processed and the
line number on which the error occurred;
for Tcl procedures, the procedure name and line number
within the procedure are recorded, and so on.
The best time to call Tcl_AddObjErrorInfo is just after
Tcl_EvalObj has returned TCL_ERROR.
In calling Tcl_AddObjErrorInfo, you may find it useful to
use the errorLine field of the interpreter (see the
Tcl_Interp manual entry for details).
Tcl_AddErrorInfo resembles Tcl_AddObjErrorInfo
but differs in initializing errorInfo from the string
value of the interpreter's result
if the error is just starting to be logged.
It does not use the result as a Tcl object
so any embedded null characters in the result
will cause information to be lost.
It also takes a conventional C string in message
instead of Tcl_AddObjErrorInfo's counted string.
The procedure Tcl_SetObjErrorCode is used to set the
errorCode variable. errorObjPtr contains a list object
built up by the caller. errorCode is set to this
value. Tcl_SetObjErrorCode is typically invoked just
before returning an error in an object command. If an error is
returned without calling Tcl_SetObjErrorCode or
Tcl_SetErrorCode the Tcl interpreter automatically sets
errorCode to NONE.
The procedure Tcl_SetErrorCode is also used to set the
errorCode variable. However, it takes one or more strings to
record instead of an object. Otherwise, it is similar to
Tcl_SetObjErrorCode in behavior.
Tcl_PosixError
sets the errorCode variable after an error in a POSIX kernel call.
It reads the value of the errno C variable and calls
Tcl_SetErrorCode to set errorCode in the POSIX format.
The caller must previously have called Tcl_SetErrno to set
errno; this is necessary on some platforms (e.g. Windows) where Tcl
is linked into an application as a shared library, or when the error
occurs in a dynamically loaded extension. See the manual entry for
Tcl_SetErrno for more information.
Tcl_PosixError returns a human-readable diagnostic message
for the error
(this is the same value that will appear as the third element
in errorCode).
It may be convenient to include this string as part of the
error message returned to the application in
the interpreter's result.
It is important to call the procedures described here rather than
setting errorInfo or errorCode directly with
Tcl_ObjSetVar2.
The reason for this is that the Tcl interpreter keeps information
about whether these procedures have been called.
For example, the first time Tcl_AddObjErrorInfo is called
for an error, it clears the existing value of errorInfo
and adds the error message in the interpreter's result to the variable
before appending message;
in subsequent calls, it just appends the new message.
When Tcl_SetErrorCode is called, it sets a flag indicating
that errorCode has been set;
this allows the Tcl interpreter to set errorCode to NONE
if it receives an error return
when Tcl_SetErrorCode hasn't been called.
If the procedure Tcl_ResetResult is called,
it clears all of the state associated with
errorInfo and errorCode
(but it doesn't actually modify the variables).
If an error had occurred, this will clear the error state to
make it appear as if no error had occurred after all.
