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

Re: netCDF3.5 test failure with Cygwin GNU gcc - nctest success (fwd)

------- Forwarded Message

Date:    Thu, 19 Apr 2001 13:00:57 -0600
From:    Russ Rew <address@hidden>
To:      "Forsberg, Carl Fredrik" <address@hidden>
Subject: Re: netCDF3.5 test failure with Cygwin GNU gcc - nctest success 

>From: "Forsberg, Carl Fredrik" <address@hidden>
>Subject: netCDF3.5 test failure with Cygwin GNU gcc
>Organization: Norwegian Polar Institute
>Keywords: 200104191239.f3JCd8L09299 netCDF make test

Carl,

> After having sent my previous e-mail, I tested the netCDF with nctest and
> received no errors and a ******success****** message.  I also tried nc_test
> (make test) and nf_test and received a *****success***** message from these
> as well, although I am not certain if anything was tested with nf_test (no
> OKs).
> 
> I then installed netCDF and the GMT version 3.4 package.  I checked some
> plots that use grids (netCDF) and found no change from previous versions.  I
> am therefore inclined to think that the t_nc program or the cmp test in the
> makefile is incompatible with Cygwin.  I have not dug into the t_nc code to
> check.  Any opinions?

Thanks for reporting back on this.  The failure in running the t_nc
test:

 cmp test.nc test_nc.sav
 test.nc test_nc.sav differ: char 1545, line 5

might be a very small difference of 1-bit in a floating-point value,
or might be something more significant.  The best way to tell would be
to run the "ncdump" utility (built as part of compiling the library)
on both netCDF files and see what the difference was, as in:

 ncdump test_nc.sav > test1.cdl
 ncdump test.nc > test2.cdl
 
(or however you send output to a file, sorry I'm not Windows-literate)
and then compare the text files test1.cdl and test2.cdl to see what
the difference is.

If this indicates only a tiny floating-point error, then you needn't
worry.  If it indicates a problem using character-valued variables,
for example, then that's more serious.

--Russ


------- End of Forwarded Message