| erract |
|
Table of contents
Procedure
ERRACT ( Get/Set Default Error Action )
SUBROUTINE ERRACT ( OP, ACTION )
Abstract
Retrieve or set the default error action.
Required_Reading
ERROR
Keywords
ERROR
Declarations
IMPLICIT NONE
CHARACTER*(*) OP
CHARACTER*(*) ACTION
Brief_I/O
VARIABLE I/O DESCRIPTION
-------- --- --------------------------------------------------
OP I Operation -- 'GET' or 'SET'
ACTION I-O Error response action
Detailed_Input
OP indicates the operation -- 'GET' or 'SET'. 'GET' means,
"Set ACTION to the current value of the error response
action."
'SET' means, "update the error response action to the
value indicated by ACTION."
OP may be in mixed case; for example,
CALL ERRACT ( 'gEt', ACTION )
will work.
ACTION when OP is 'SET', ACTION is an input argument. It
takes the values, 'ABORT', 'IGNORE',
'REPORT', 'RETURN', and 'DEFAULT'.
Please read the "required reading" file if you
haven't already done so!
Briefly, the meanings of the error response
choices are as follows:
1. 'ABORT' -- When an error is detected by a
SPICELIB routine, or when
ANY routine signals detection
of an error via a call to SIGERR, the
toolkit will output any error messages that
it has been enabled to output (see ERRPRT
and ERRDEV also ), and then execute a
FORTRAN STOP statement.
2. 'REPORT' -- In this mode, the toolkit does
NOT abort when errors are detected.
When SIGERR is called to report
an error, all error messages that the toolkit
is enabled to output will be sent to the
designated error output device. Similarly,
a call to SETMSG will result in the long
error message being output, if the toolkit
is enabled to output it.
3. 'RETURN' -- In this mode, the toolkit also
does NOT abort when errors are
detected. Instead, error messages
are output if the toolkit is enabled to do
so, and subsequently, ALL TOOLKIT ROUTINES
RETURN IMMEDIATELY UPON ENTRY until the
error status is reset via a call to RESET.
(No, RESET itself doesn't return on entry).
Resetting the error status will cause the
toolkit routines to resume their normal
execution threads.
4. 'IGNORE' -- The toolkit will not take any
action in response to errors;
calls to SIGERR will have no
effect.
5. 'DEFAULT' -- This mode is the same as 'ABORT',
except that an additional error
message is output. The additional
message informs the user that the
error response action can be
modified, and refers to documentation
of the error handling feature.
ACTION may be in mixed case; for example,
CALL ERRACT ( 'SET', 'igNORe' )
will work.
Detailed_Output
ACTION when OP is 'GET', ACTION is the current error response
action. Possible values are: 'ABORT', 'REPORT',
'RETURN', and 'IGNORE'. See "Detailed Input"
for descriptions of the meanings of these values.
ACTION is not an output unless OP is 'GET'.
Parameters
None.
Exceptions
1) If an invalid value of the operation OP is supplied, the error
SPICE(INVALIDOPERATION) is signaled.
2) If OP is 'SET' and the input argument ACTION does not indicate
a valid error handling action, the error SPICE(INVALIDACTION)
is signaled.
Files
None.
Particulars
This routine is part of the SPICELIB error handling mechanism.
First of all, please read the ``required reading'' file.
The information below will make a lot more sense if you do.
Here is a brief discussion of how to use this routine.
If you are a user, you will probably be interested
in only the 'SET' operation (as far as this routine is
concerned, ok?). As indicated in the "detailed
input" section above, the choices for ACTION are
'ABORT', 'REPORT', 'RETURN', 'IGNORE', and 'DEFAULT'. These
choices control the way the toolkit behaves when an
error is detected. The toolkit thinks an error has
been detected when SIGERR is called.
1. 'ABORT' In this mode, the toolkit sends error messages
to the error output device and then stops.
This is the default mode. It is probably
the one to choose for running non-interactive programs.
You may also wish to use this for programs which
have many bugs, or in other cases where continued
operation following detection of an error isn't useful.
2. 'REPORT' In this mode, the toolkit sends error messages
to the error output device and keeps going. This mode
may be useful if you are debugging a large program,
since you can get more information from a single test run.
You will probably want to use ERRDEV to indicate a file
where your error messages should be sent.
3. 'RETURN' In this mode, the toolkit also sends error messages
to the error output device and "keeps going". But
instead of following their normal execution threads,
the toolkit routines will simply return immediately upon
entry, once an error has been detected.
The availability of this feature makes it safe to call
multiple toolkit routines without checking the error
status after each one returns; if one routine detects
an error, subsequent calls to toolkit routines will have
no effect; therefore, no crash will occur. The error
messages set by the routine which detected the error
will remain available for retrieval by GETMSG.
4. 'IGNORE' This mode can be dangerous! It is best
used when running a program whose behavior you
understand well, in cases where you wish to suppress
annoying messages. BUT, if an unexpected error
occurs, you won't hear about it from anyone, except
possibly your run-time system.
5. 'DEFAULT' As the name suggests, this is the default
error handling mode. The error handling mechanism
starts out in this mode when a program using the
toolkit is run, and the mode remains 'DEFAULT' until
it is changed via a call to this routine.
This mode is the same as 'ABORT',
except that an additional error message is output.
The additional message informs the user that the
error response action can be modified, and refers
to documentation of the error handling feature.
NOTE:
By default, error messages are printed to the screen
when errors are detected. You may want to send them
to a different output device, or choose a subset to
output. Use the routines ERRDEV and ERRPRT to choose
the output device and select the messages to output,
respectively.
You can also suppress the automatic output of messages
and retrieve them directly in your own program. GETMSG
can be used for this. To make sure that the messages
retrieved correspond to the FIRST error that occurred,
use 'RETURN' mode. In 'REPORT' mode, new messages
overwrite old ones in the SPICELIB message storage
area, so GETMSG will get the messages from the LATEST
error that occurred.
Examples
1. Setting up 'ABORT' mode:
C
C We wish to have our program abort if an error
C is detected. But instead of having the error
C messages printed on the screen, we want them
C to be written to the file, ERROR_LOG.DAT
C (This is valid VAX/VMS file name; syntax
C on your system may be different ).
C
C We want to see all of the messages, so we
C call ERRPRT, using the 'ALL' option.
C
C Finally, we call ERRACT to set the action to 'ABORT':
C
CALL ERRDEV ( 'SET', 'ERROR_LOG.DAT' )
CALL ERRPRT ( 'SET', 'ALL' )
CALL ERRACT ( 'SET', 'ABORT' )
2. Setting up 'REPORT' mode:
C
C This is the same thing as before, except
C that the argument supplied to ERRACT
C is different.
C
CALL ERRDEV ( 'SET', 'ERROR_LOG.DAT' )
CALL ERRPRT ( 'SET', 'ALL' )
CALL ERRACT ( 'SET', 'REPORT' )
3. Setting up 'RETURN' mode: This is the same
as example #2, except that the ERRACT call becomes:
CALL ERRACT ( 'SET', 'RETURN' )
4. Setting up 'IGNORE' mode:
C In this case, we aren't going to have
C ANY error messages (unless the call
C to ERRACT itself fails), so we don't
C really need to call ERRPRT and ERRDEV.
C (If the call to ERRACT DOES fail, which
C it can do only if we misspell "IGNORE,"
C the resulting error messages will go to
C the screen).
CALL ERRACT ( 'SET', 'IGNORE' )
Restrictions
None.
Literature_References
None.
Author_and_Institution
N.J. Bachman (JPL)
J. Diaz del Rio (ODC Space)
K.R. Gehringer (JPL)
W.L. Taber (JPL)
Version
SPICELIB Version 2.1.0, 12-AUG-2021 (JDR)
Added IMPLICIT NONE statement.
Edited the header to comply with NAIF standard. $Exceptions
section has been completely updated to provide only the list
of exceptions. Additional information provided there has been
moved to $Particulars.
SPICELIB Version 2.0.0, 22-APR-1996 (KRG)
This subroutine has been modified in an attempt to improve
the general performance of the SPICELIB error handling
mechanism. The specific modification has been to change the
type of the error action passed to PUTACT from a short
character string to an integer. This change is backwardly
incompatible because the type of the input argument has
changed. This should pose no difficulties because PUTACT is a
private subroutine used by the error handling system, and
hence isolated from direct use.
SPICELIB Version 1.0.1, 10-MAR-1992 (WLT)
Comment section for permuted index source lines was added
following the header.
SPICELIB Version 1.0.0, 31-JAN-1990 (NJB)
|
Fri Dec 31 18:36:20 2021