[udf-dlm-develop] Found it!!!

[Harald F. <hf...@ap...>]

The "nasty problem" became a "Question of Honor" and I traced it down to the 
source. The whole problem occurs, if you open a UDF file and don't(!) read to 
the end of it, so that udf_eof(fh) does not become 1.

I have attached an IDL program, which shows this behavior. In the first round I 
open three files and read all data to their end. In the second round I open the 
files, but read only one record at the beginning of the file. And then the 
sequence crashes, when I try to access a file with later data. The same happens 
if I don't read anything but only open and close the files.

This is my log of the attached program

IDL> test_udf_dlm
% Loaded DLM: UDF.
Starttime, stoptime
    1999     337       7      21      11     530
    1999     337       7      27      21     538
Starttime, stoptime
    1999     186       1      42      16     686
    1999     186       2      56      26     798
Starttime, stoptime
    1999     189      12      54       5     447
    1999     189      18      52      15     730
   
We repeat the previous sequence but read only one record
   
Starttime, stoptime
    1999     337       7      21      11     530
    1999     337       7      21      11     530
Starttime, stoptime
    1999     186       1      42      16     686
    1999     186       1      42      16     686
Starttime, stoptime
    1999     189      12      54       5     447
    1999     189      12      54       5     447
% UDF_OPEN: no data for this time period
% Execution halted at:  TEST_UDF_DLM       70 
/disks/sprite/disk1/hfrey/idl/image/dlm/udf-dlm-0.52/test_udf_dlm.
  pro
%                       $MAIN$                 
% Program caused arithmetic error: Floating overflow

Harald

=========================================================
Harald U. Frey 
Space Sciences Lab	   phone: 510-643-3323
University of California   fax:   510-643-2624      
Berkeley, CA 94720-7450    email: hf...@ss...