[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #DNU-747889]: NetCDF memory problem?

> I had been using NetCDF (a version from about 3 years ago) that another
> co-worker had gotten to run on Windows.  I saw that when I was trying to get
> data with an index > 1024, the values that were returned by NetCDF were
> fixed at a constant value; i.e., as I increased the coord index on the
> second parameter past 1024, I always got the same data value back for a
> given parameter.
>
>
>
> I am finally getting very desperate to have this fixed, so today, I decided
> to see if the latest version of netCDF worked any better, so I downloaded
> the latest version, and loaded the .NET solution into Visual Studio 2008.
> After MUCH angst (__int64_t vs _int64, and __uint8 vs unsigned __int8, to
> mention but a few), I got it successfully running, starting from the .NET
> solution that comes with the download.  Unfortunately, I get the same
> result, and the data appears to be invalid.
>
>
>
> For example, I am using it on the NCEP pressure data, for 2005, using a date
> of 2005/09/14, which corresponds to a time coord value of 1025.  That
> doesn't work (all dates past and including 2005/09/14 return the same data),
> but if I go to 2005/09/13, the corresponding coord value is 1020, and the
> data changes with each variation in system time.
>
>
>
> To summarize, if I put the "ncvarget1(ncid, 4, corner_LR, (void *)
> &short_hi);" and step thru the time dimension and create a carpet plot, the
> values vary appropriately until the time coordinate gets larger than 1024.
>
>
>
> Question:  Does the .NET solution that's included in the current 4.0
> download run anything that is v.4.0, or is it still compiling the old 3.6.3
> version?
>
>
>
> Any ideas how to work around this?  Is there some global variable I need to
> set to get the netCDF software work with larger files?
>
>
>
> Daniel L. Oltrogge, Senior Analyst & CEO, 1Earth Research, LLC
>
> PO Box 51412, Colo. Springs, CO 80949; www.1EarthResearch.com
>
> Voice: 719-660-5142; E-mail: address@hidden
>
>

Howdy Dan!

Sorry to hear about your troubles with Windows. It is indeed a problem. If you
could send me your solution file, I can make it available.

The solution file distributed with the release works great, with .NET 2003. It
is irritating that Microsoft so often feels the need to change things such that
all the configuration and code files break. (On Unix systems we are still using
stuff that worked fine 20 years ago.)

On the other hand, I do recognize that Windows is what people want to use, and
so I have plans to get .NET 2008 and "upgrade" my solution file so that it
builds there.

Did you get the netCDF tests to run successfully? If so, you can be pretty
confident that netCDF is working pretty well.

As I understand your problem, you think the data are there in the file, but
your program is giving you bad values. Do you have ncdump.exe? (If not, you can
build it.)

ncdump will show you the contents of the file to confirm that it contains what
you think it does. It will also check that the netcdf library is working
correctly in this case. If you run ncdump -h and send me the output I can help
with your problem. I am not familiar with the NCEP data, sorry.

To answer another of your questions, the solution file distributed with the
release does not include netCDF-4 features. I am hoping to add those when I get
the .NET 2008 solution file working.

Your problems sound as if they are related to the index information in your
data. If your netcdf library is working for index values below 1024, it is
likely that it is working just fine, and you are asking for something that is
not there...

Anyway, please send me the ncdump -h output of the file, or the file itself if
necessary, and we will see...

Thanks,

Ed

Ticket Details
===================
Ticket ID: DNU-747889
Department: Support netCDF
Priority: Normal
Status: Open