NC_open

Russ

This is a possible bug report regarding


https://sourceforge.net/projects/nco/forums/forum/9829/topic/8755743



I am using this command

ncks -H -d lat,32.,28. -d lon,245.5.,248.7. 
http://pcmdi7.llnl.gov/thredds/dodsC/cmip5_css02_data/cmip5/output1/MPI-M/MPI-ESM-MR/historical/mon/ocean/Omon/r1i1p1/tos/1/tos_Omon_MPI-ESM-MR_historical_r1i1p1_200001-200512.nc 
test.nc


On the function NC_open in dfile.c


There is a variable named
version

that is not initialized before being used with this case
this causes my compiler to exit (using Windows , Visual Studio)




int
NC_open(const char *path, int cmode,
int basepe, size_t *chunksizehintp,
       int useparallel, void* mpi_info,
       int *ncidp)
{
  int stat = NC_NOERR;
  NC* ncp = NULL;
  NC_Dispatch* dispatcher = NULL;
  /* Need two pieces of information for now */
  int model = 0;
  int isurl = 0;
  int version;
  enum FileType filetype;



The error happens because of this use of "version"

else if(model & NC_DISPATCH_NC3) {
     cmode &= ~NC_NETCDF4; /* must be netcdf-3 */
     if(version == 2) cmode |= NC_64BIT_OFFSET;



I tried to initialize

version = 0

to further debug this

/* Pass to OC */
   ocstat = oc_open(dapcomm->oc.urltext,&dapcomm->oc.conn);

this call returns no error

these calls

ocerr = ocfetch(state,constraint,dxdkind,flags,&root);

stat = readDDS(state,tree);

return an error code of -13


this call


/* fetch and build the unconstrained DDS for use as
      template */
   ncstat = fetchtemplatemetadata3(dapcomm);

returns a error value of -68

I am not very familiar with the OC return codes, but this seems to be an 
issue with the URL or file , not a netCDF issue ?

Thanks

Pedro


------------
address@hidden
Department of Earth System Science
3200 Croul Hall
University of California, Irvine
Irvine, CA 92697-3100




----- Original Message ----- 
From: "Russ Rew" <address@hidden>
To: "Pedro Vicente" <address@hidden>
Cc: <address@hidden>; "Charlie Zender" <address@hidden>
Sent: Saturday, April 13, 2013 10:39 AM
Subject: Re: [netCDF #SHH-257980]: Re: [netcdfgroup] Dimensions IDs


> Pedro,
>
> Here's the Jira ticket for the bug you reported:
>
>  https://bugtracking.unidata.ucar.edu/browse/NCF-247
>
> A CDL file that triggers the bug is now only 13 lines, so I hope to be
> able to fix this one soon ...
>
> --Russ