Fix your code to initialize your variables. The p-code compiler does that
for you -- arguably this is a bug, because the C compiler definitely doesn't.
This is probably the biggest unresolvable difference between the p-code and
c-code run-time systems.
> There is something wrong here because this code worked on c4gl
> version 4 4gl on AT&T Unix SysV3.2 with version 4 SE.
Pure coincidence -- there just happened to be a zero byte at the correct
location. You were unbelievably lucky. Well, maybe not unbelievably, but
definitely lucky.
Yours,
Jonathan Leffler (johnl@informix.com) #include <disclaimer.h>
}From: john@gagme.wwa.com (pinoy_ako)
}Date: 10 Mar 1995 01:29:36 -0600
}X-Informix-List-Id: <news.12108>
}
} We've go a program fragment like this: (clipped to minimum so
} syntax isn't correct)
}
} FUNCTION hello(var1, var2, var3)
} ...
} DEFINE var3 like sometable.somecol
} .
} IF var3 IS NULL THEN
} # do this thing
} ELSE
} # do another thing
} END IF
}
} The function is called like this:
}
} CALL hello(param1, param2, param3)
}
} The problem occurs when param3 is not set to any value (only defined
} like this 'DEFINE param3 CHAR(5)', no param3 = "" or initialize param3
} to null or param3 = "value"). THe program goes into the 'do other
} thing' part instead of the 'do this thing'. I tried adding this code
} to the function hello, above the IF statement so that I can trace
} the variable var3 is being set to:
}
} ERROR "this is var3: ", var3
}
} It showed me the following everytime I ran it:
}
} this is var3: >`@
}
} There is something wrong here because this code worked on c4gl version
} 4 4gl on AT&T Unix SysV3.2 with version 4 SE. Now that we have version
} 4.12 4gl, C-compiler version on SCO unix 3.2.4 with 5.02 SE, it just
} doesn't initialize the variable correctly. Can any one suggest a
} solution? I can't just initialize var3 to NULL because param3 might
} have a value. This is what I have already tried (but failed)