CM> MK> Be sure to initialize InstanceData.ptrUserProfile to NULL and
  > MK> reinitialize to NULL if what it points at becomes invalid (like if you
  > MK> free it).

CM>Actually, it turned out to be a very strange problem.  I
  >was calling a function long before this code is run, that
  >assignes a character to a variable defined as
  >        char Temp;

CM>Well, if I simply recode this to be:
  >        char Temp[255];

CM>The problem with sprintf many, many lines, of code, and
  >several functions later, disappears.

CM>Could this be a C/SET bug, or just a very odd artifact of what I was doing?

Actually, it is a sign of stray variables.  Now I know that OS/2 tries
it's best to stop those, but they do still happen.  If all you are doing
is assigning a char value to the Temp, then you should not need the 255,
infact that would be wasting 254 bytes of memory!  It sounds to me that
perhaps a pointer to one of your functions is being over-written when
Temp is a single char, but when Temp is 255 characters long, its
protected.

Here's what I see:

legend: [ ] <-space allocated for Temp
         *  <-area where stray data is being written

char Temp:       [ ]             **
char Temp[255]:  [               **         ]

See, instead of allowing your stray-written garbage to overwrite a
function pointer or some similarly dangerous piece of data, you allow it
to fall within a region of unused space (since you are only using the
first byte of Temp).

Suggestion, leave Temp as Temp[255], and debug your program one function
at a time.  Watch for any change (other than those that should take
place) in Temp.  When you find something changed, you've found your
offending piece of code.

Michael Douglass
___
 .Mike's Mail Internet:   MICHAEL.DOUGLASS{at}LCHANCE.SAT.TX.US
  
--- Maximus/2 2.01wb