Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!samsung!emory!gatech!prism!gt4586c
From: gt4586c@prism.gatech.EDU (WILLETT,THOMAS CARTER)
Newsgroups: comp.sys.mac.programmer
Subject: Strange Think Pascal Problem - Help?!
Keywords: compile options problem
Message-ID: <15053@hydra.gatech.EDU>
Date: 12 Oct 90 04:57:56 GMT
Organization: Georgia Institute of Technology
Lines: 22

I have been writing a number crunching program, and in the process I have 
come across a most perplexing problem.  I am using Think Pascal 3.0.  I am
reading a data file into a buffer, and then reading off chunks of the buffer
to get the numbers.  My routine works fine when I have the compile options
set to generate '020/030 machine code and making calls to the FPU, with
ELEMS881 = TRUE.  However, when I change it to generate code for the SE, 
it goes whacky.  The data gets read into my arrays, and when I plot the
data it has the correct shape, but the magnitudes are reduced by about 65536,
or 2 to the 16th power.  This suggests that for some reason the first 16 bits
of each number i try to read off are being zeroed out, effectively bit-
shifting-right sixteen places.  I am trying to read extended precision
(80 bit) numbers out of the file buffer.  I find it truly bizarre that it
works just fine with the MacII version but bombs with the SE version.  No
code is changed - only the compile options.  Does anybody have any idea
what might be going wrong?  Thanks.


-- 
thomas willett 
Georgia Institute of Technology, Atlanta  
gt4586c@prism.gatech.edu
"Violence is the last refuge of the incompetent." - Salvor Hardin (Foundation)