On Tue, 7 Oct 1997, Stan Sieler wrote:

> Srivathsa asks:
 [...edit...]
> > The problem that I have is that the compilation of MPE 4.0 source codes on MPE 5.0
> > goes through well,  but the product doesn't run at all. When  the MPE 4.0 product
> > executable is put in the appropriate groups/account in MPE 5.0, it runs perfectly.
> > The communicator for MPE 5.0 says that any application written for MPE XL works
> > fine with MPE/iX without any change. I am attaching herewith a copy of the spool
> > output when I start running the product. Can you please tell me what the problem
> > could be?
 [...edit...]
> 1) My understanding is that HP Sockets (for the 3000) is supposed to be compiled
>    on HP-UX, and then *linked* on a 3000.  If true, then by changing the build
>    process to compile on the 3000, you're getting different compilers, and
>    introducing new variables into the equation.  (Perchance, were you unable
>    to compile on the 9000 because of 10.x changes?)

 All of my socket programming for the 3000 has been done entirely on the
3000.  Compiled with CCXL and linked either to an XL or an executable
binary.

> >  :run smd.pub.hpskts
> >
> >  **** Data memory protection trap (TRAPS 68).
> >
> >  ABORT: SMD.PUB.HPSKTS
> >  NM PROG  5f8.0004d778 SpGet_Next_Msg+$70

 Generally... when I get an error like this it's because of something
stupid I've done with string manipulation.  I tend to forget that MPE/iX
is a lot more stringent than Un*x when it comes to variable memory
allocation.  I, sometimes, get lazy and forget to malloc() space for my
variables and it eventually goes boom.

 Srivathsa, can you provide the code segment from the routine
SpGet_Next_Msg()???  I'm assuming this is C source?

I hope that helps,
Jim
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
James R. Alton                        phone: (905) 575-2281
Technical Support Specialist            Fax: (905) 575-2302
Mohawk College                       mailto:[log in to unmask]