I've run up against what I believe may be a bug importing native DX
files when certain components refer to dx objects in different files.

Let me describe what I'm doing.  I have a (time) sequence of scalar
fields that share a common fixed unstructured tetrahedral mesh.  The
positions and connections arrays (data and headers) are written
to one file.  The field data arrays (connection dependent) are written
to a second file, where the field objects are defined; their position
and connection components refer to the array objects in the first
file.  A series object that collects all the field objects together
is also defined in the second file.  Both files consist of a header
section followed by a (lsb) binary data section.  (I actually have
six different scalar field sequences, and was originally writing
everything to a single file, but the files were becoming so enormous
that I've gone to writing the mesh and field sequences to different
files to make things more managable.)

If the size of the data arrays are sufficiently large and/or there are
sufficiently many members in the series, then importing the series
will fail with the error, "Import: Invalid data: bad or missing
series member object/file <file> line <n>".  There appears to be
sufficient memory; if everything is written to a single file there
is no problem importing the series.  A series with 30 members, and
fields over a mesh of 660K tetrahedra is sufficiently large to make
the import fail.

I'd post the dx files if they were small (unfortunately they're over
90MB :-).  However it seems pretty easy to trigger the error; I've
written a test program to write the dx files with arbitrary data.
It would seem that repeated references to objects in another file
is overflowing internal tables or something.

Any thoughts about how to proceed on tracking down the problem?

Thanks,
  Neil

PS: I'm running 4.2 on linux.
-- 
Neil N. Carlson                    Motorola, Los Alamos Research Park
Motorola Labs / PSRL               4200 W. Jemez Road, Suite 300
Computational Nanoscience Group    Los Alamos, New Mexico 87544

[EMAIL PROTECTED]           Fax: (505) 663-5150
Voice: (505) 663-5106              Pager: [EMAIL PROTECTED] or 888-946-2817

Reply via email to