| Re: READU on IRX and Win95 [message #16336] |
Tue, 20 July 1999 00:00 |
Chris Jengo
Messages: 13
Registered: July 1999
|
Junior Member |
|
|
> When I use READU on the same file in Win95 and IRIX, I get different
> results.
> Is this possible? How would I read the file on a Win95 system?
Looks like a byte swapping problem. No matter how many times this happens
to me, it always takes me a few minutes to figure out what the heck is going
on! In ENVI I just switch it in the header, but in IDL it looks like
BYTEORDER will do the trick (going from network (IEEE) to Host (Intel), use
the XDRTOF keyword).
Chris
______________________
Chris Jengo
Imaging Scientist
Earth Satellite Corporation
(301)231-0660
cjengo@earthsat.com
|
|
|
|
| Re: READU on IRX and Win95 [message #16343 is a reply to message #16336] |
Tue, 20 July 1999 00:00  |
Liam Gumley
Messages: 473
Registered: November 1994
|
Senior Member |
|
|
Paul O Storaasli wrote:
> When I use READU on the same file in Win95 and IRIX, I get different
> results. Is this possible?
Oh yes. A quote from the link referenced below:
"Little and Big-Endian machines
------------------------------
Another practical complication arises when you store any multi-byte
entity in memory.
Computer memory is referenced by addresses that are positive integers.
It is 'natural' to store numbers with the LEAST SIGNIFICANT BYTE coming
before the MOST SIGNIFICANT BYTE in the computer memory, however
computer designers prefer sometimes to use a reversed order version of
the representation.
The 'natural' order, where less significant binary digits comes before
more significant digits in memory is called LITTLE-ENDIAN, many vendors
like IBM, CRAY and Sun preferred the reverse order that of course is
called BIG-ENDIAN."
How would I read the file on a Win95 system?
> I am trying to read a binary file of 32-bit floats with the following
> code:
>
> function readfile
> ; This function will read the data in 'D:\data\thunderb.653'
> ; as an array of floating point (32-bit) numbers. It assumes
> ; the array has 716 colums, and calculates the number of rows.
> ; It returns the array.
>
> file='thunderb.653'
>
> openr,LUN,file,/Get_Lun
> result=fstat(LUN)
> nbytes=result.size
> nx=716L
> ny=nbytes / 4L /nx
> image=fltarr(nx,ny)
> readu,LUN,image
> free_lun, LUN
>
> return, image
> end
>
> The results I get are quite different:
>
> On Win95
> IDL> print,a(1:5,1:5)
> -6.32813e-023-4.27821e+008-6.32846e-023-4.27903e+008 1.08637e+024
> 9.04917e-041-4.27821e+008 8.09951e-043 3.67981e-042 4.16997e-008
> 9.24857e-044-6.32813e-023-4.27838e+008-6.33169e-023-6.33524e -023
> 1.08622e+024 8.09951e-043 4.16942e-008-6.33718e-023-4.28427e+008
> -4.27903e+008 3.67981e-042 1.44418e-041-4.28427e+008-6.34203e-023
> % Program caused arithmetic error: Floating underflow
>
> On an IRIX system:
> IDL> print,a(1:5,1:5)
> 32.4000 32.2000 32.9000 34.7000 35.1000
> 31.5000 32.2000 32.5000 34.5000 38.3000
> 32.0000 32.4000 32.7000 37.9000 43.4000
> 33.1000 32.5000 36.8000 46.4000 50.7000
> 34.7000 34.5000 42.0000 50.7000 53.9000
Under Win95, try typing
IDL> print, swap_endian(a[1:5,1:5])
If the data file is created under Irix in IDL, add the /XDR keyword to
your OPENW and OPENR statements. This will cause the data to be written
and read in a format which is portable between different IDL machine
architectures.
If the data file is created under Irix in some other application, add
the keyword /SWAP_IF_LITTLE_ENDIAN to the OPENR statement, which will
cause the data to be swapped if the file is opened on a little-endian
system, like a PC.
For a full explanation of this behavior (which is not specific to IDL),
check out
http://metalab.unc.edu/pub/languages/fortran/ch4-3.html
Cheers,
Liam.
--
Liam E. Gumley
Space Science and Engineering Center, UW-Madison
http://cimss.ssec.wisc.edu/~gumley
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
