Re: NetCDF Digest - Vol 1 : Issue 76

• To: netcdfgroup@xxxxxxxxxxxxxxxx
• Subject: Re: NetCDF Digest - Vol 1 : Issue 76
• From: Harvey Davies <hld@xxxxxxxxxxxx>
• Date: Mon, 26 Jul 1993 09:38:36 +1200 (EST)

> 07/22 David Simas -- "imap" argument of netCDF function "ncvargetg"
> 
> ------------------------------------------------------------
> 
> Subject: "imap" argument of netCDF function "ncvargetg"
> From:    dsimas@xxxxxxxxxxxx (David Simas)
> Date:    Thu, 22 Jul 93 11:53:44 PDT
> 
> I'm using the function "ncvargetg" to read a two-dimensional array of
> variables from a netCDF file.  Under certain circumstances, I'd like to
> invert th order of the columns and/or rows of that array.  I assume that
> can be done by manipulating the "imap" argument
>  to "ncvargetg", but I'm not sure how.  Neither the netCDF manual's
> explanation of "imap" and "ncvargetg" nor the example of their 
> use given there make any sense.  If anyone knows how to use "ncvargetg"
> and "imap", I'd very much like to hear from him or her.

I also had trouble using ncvargetg & ncvarputg. There was a serious bug in
original release of version 2.3. This has now been fixed.

There were also errors in User's Guide. In examples on pages 87 & 101 the
final argument of final line (calling ncvarputg/ncvargetg) should be
(void *) &data[0][0][0].rh_val
rather than
(void *) data

The basic idea is that the address of element [ i[0], i[1], i[2], ... ]
is given by
value + imap[0] * i[0] + imap[1] * i[1] + imap[2] * i[2] + ...

Note that subscript i[d] takes the values
start[d]
start[d] + stride[d]
start[d] + 2 * stride[d]
...
start[d] + (count[d]-1) * stride[d]

I hope this is clear. If not I could provide an example showing how to 
transpose a matrix. There is a need for such a simple example in the
User's Guide.


Harvey Davies,
CSIRO Division of Atmospheric Research,    Voice: +61 3 586 7574
Private Bag No. 1, Mordialloc,               Fax: +61 3 586 7600
Victoria 3195,  Australia               Internet: hld@xxxxxxxxxxxx