zhou@brazil.psych.purdue.edu (Albert Zhou) (01/01/91)
Whenever a floating point overflow error is encountered, the program
will unconditionally stop in Turbo Pascal. Is there a way of deactivating it
so as to let the programmer handle this type of error?
As I understand, there is no compiler derective that does it. Just
like {$I-} that enables the programmer to handle I/O errors.
Any clue will be greatly appreciated.dslg0849@uxa.cso.uiuc.edu (Daniel S. Lewart) (01/01/91)
zhou@brazil.psych.purdue.edu (Albert Zhou) writes: > Whenever a floating point overflow error is encountered, the program > will unconditionally stop in Turbo Pascal. Is there a way of deactivating it > so as to let the programmer handle this type of error? > As I understand, there is no compiler derective that does it. Just > like {$I-} that enables the programmer to handle I/O errors. > Any clue will be greatly appreciated. Yes, you can point the ExitProc pointer variable to your exit procedure. It is discussed in the manual. Daniel Lewart d-lewart@uiuc.edu
zhou@brazil.psych.purdue.edu (Albert Zhou) (01/01/91)
In article <1991Jan1.033432.11437@ux1.cso.uiuc.edu> dslg0849@uxa.cso.uiuc.edu (Daniel S. Lewart) writes: > >Yes, you can point the ExitProc pointer variable to your exit procedure. >It is discussed in the manual. > It didn't turn out to be working well. Following is the code that gets stuck after executing my own error processing subroutines: procedure error; begin writeln('Error!'); (***** gets stuck here ******) end; const x : real = 3000; begin exitproc := @error; x := exp(x); (**** exp(3000) should result in overflow ****) end.
guillory@storm.dnet.nasa.gov (Anthony R. Guillory (904)) (01/02/91)
You need to compile your "procedure error" with {$F+}, i.e,
{$F+} Procedure Error; {$F-}
[your code]
You should also save the address of TP original exitProc by using
SaveExitProc := exitproc;
exitproc := @error;
Then at the end of your "procedure error" set exitproc := SaveExitProc
This is all documented in the manual, thelp, and most 3rd party books.
Anthony
Guillory@storm.dnet.nasa.govzhou@brazil.psych.purdue.edu (Albert Zhou) (01/03/91)
In article <25363@adm.brl.mil> guillory@storm.dnet.nasa.gov (Anthony R. Guillory (904)) writes: >You need to compile your "procedure error" with {$F+}, i.e, >{$F+} Procedure Error; {$F-} If I set {$F-} then the TP error will not be generated, but the program gets stuck. However, if I set {$F+}, the procedure error will be executed normally, but a TP error will still be generated. It's really weird. It happens to me once in a while when everything looks right and the program doesn't do what it's supposed to do.
guillory@storm.dnet.nasa.gov (Anthony R. Guillory (904)) (01/03/91)
Here's a small excerpt from an run-time error handling routine I wrote.
I don't know if this will help, but here it is.
{$F+} Procedure RuntimeErrorHandler; {$F-}
Begin
If (ExitCode <> 0) AND (ErrorAddr <> NIL)
Then Begin
[Print "My" Error Message]
ErrorAddr := NIL; { Cancel Runtime Error }
ExitCode := 0;
End { Then };
ExitProc := SavedExitProc;
End { RuntimeErrorHandler };
Begin
SavedExitProc := ExitProc; { Save ExitProc pointer }
ExitProc := @RuntimeErrorHandler;
End. { ExitProc }
_______________________________________________________________________________
Anthony R. Guillory
INTERNET: Guillory@Storm.dnet.nasa.gov
UUCP: {ames,east,decwrl,ucbvax,uunet}!storm.dnet.nasa.gov!guillory
CompuServe: 73427,2515
GEnie: A.GUILLORY
FidoNet: Anthony Guillory @ 1:3605/46
DECNET/SPAN: STORM::GUILLORY or 723::GUILLORY
o If practice makes perfect, and nobody's perfect, why practice?
o The solution of this problem is trival and is left as an exercise for
the reader.
o Recursive,adj.; see recursive.
_______________________________________________________________________________ts@uwasa.fi (Timo Salmi) (01/04/91)
In article <11637@j.cc.purdue.edu> zhou@brazil.psych.purdue.edu (Albert Zhou) writes: > > Whenever a floating point overflow error is encountered, the program >will unconditionally stop in Turbo Pascal. Is there a way of deactivating it >so as to let the programmer handle this type of error? > As I understand, there is no compiler derective that does it. Just >like {$I-} that enables the programmer to handle I/O errors. > Any clue will be greatly appreciated. 25. ***** Q: How can I trap a runtime error? A: What you are probably asking for is a method writing a program termination routine of your own. To do this, you have to replace Turbo Pascal's ExitProc with your own customized exec procedure. Several Turbo Pascal text books show ho to do this. See eg Tom Swan (1989), Mastering Turbo Pascal 5.5, Third edition, Hayden Books, pp. 440-454; Michael Yester (1989), Using Turbo Pascal, Que, pp. 376-382; Stephen O'Brien (1988), Turbo Pascal, Advanced Programmer's Guide, pp. 28-30; Tom Rugg & Phil Feldman (1989), Turbo Pascal Programmer's Toolkit, Que, pp. 510-515. Here is an example var OldExitProcAddress : Pointer; x : real; {$F+} procedure MyExitProcedure; {$F-} begin if ErrorAddr <> nil then begin writeln ('Runtime error number ', ExitCode, ' has occurred'); writeln ('The error address in decimal is ', Seg(ErrorAddr^):5,':',Ofs(ErrorAddr^):5); writeln ('That''s all folks, bye bye'); ErrorAddr := nil; ExitCode := 0; end; {... Restore the pointer to the original exit procedure ...} ExitProc := OldExitProcAddress; end; (* MyExitProcedure *) (* Main *) begin OldExitProcAddress := ExitProc; ExitProc := @MyExitProcedure; x := 7.0; writeln (1.0/x); x := 0.0; writeln (1.0/x); {The trap} x := 7.0; writeln (4.0/x); {We won't get this far} end. ................................................................... Prof. Timo Salmi (Moderating at anon. ftp site 128.214.12.3) School of Business Studies, University of Vaasa, SF-65101, Finland Internet: ts@chyde.uwasa.fi Funet: gado::salmi Bitnet: salmi@finfun