nc_test status

Robert Latham robl at mcs.anl.gov
Mon Mar 10 17:17:43 CST 2003


On Mon, Mar 10, 2003 at 11:12:56AM -0600, Robert Latham wrote:
> Total number of failures: 41717

i'm an idiot.  i cleaned out the directory before running that test,
including the .nc file.  

it now passes a lot of tests, but does fail the ncmpi_get_var_short
test:  "test_ncmpi_get_var_short" is an m4-generated test, so it's a
bit of a pain to debug... i'm comparing ncmpi_get_var_short with
nc_get_var_short: it looks like we don't handle scalar variables
correctly?  We get a seg fault here: 

	nelems *= varp->shape[0];

and serial netcdf does this:

	if(varp->ndims == 0) /* scalar variable */
        {       
                const size_t zed = 0;
                return( getNCv_short(ncp, varp, &zed, 1, value) );
        }


I just re-read some recent messages:  Jianwei, this is exactly like
the problem you fixed for ncmpi_put_vara_text_all:  the variables have
null 'shape' and 'dsizes' arrays.  

thanks
==rob

-- 
Rob Latham
Mathematics and Computer Science Division    A215 0178 EA2D B059 8CDF
Argonne National Labs, IL USA                B29D F333 664A 4280 315B



More information about the parallel-netcdf mailing list