Re: 970624: Ask for help on nf_test with win32_install--jchiang

>
> Hi, Glenn
>
> Thanks for the instruction. After reading your e-mail,
> I tried the nf_test again according to your suggestion,
> "Change X_REAL_MAX at line 123 in nf_test/tests.inc to a
> Bigger value(original value=3.4028234663852886e+37), say
> 3.40282340e+38."  Unfortunately, this try failed at the
> same place mentioned at the previous mail. However, I
> have made several tries by changing X_REAL_MAX from
> 3.40282340e+36 to 3.40282347e+38, but all the tries failed
> in the same fashion.
> As you said the issue should not keep user from using
> the library, if the other tests work okay. When I was
> involving with netCDF-3.3 remaining tests were working
> properly except the subject test, nf_test. But, this time
> I got a new trouble in fortran/ftest which you mentioned
> especially. The following is the runtime message of fortran/ftest:
> ================ start of message =====================
> ** Testing nccre ...
> ** Testing ncddef ...
> ** Testing ncvdef ...
> ** Testing ncapt, ncaptc ...NCAPT: : Unknown error
> NCAPT: : Unknown error
> NCAPT: : Unknown error
> NCAPT: : Unknown error
>
> ** Testing ncclos ...
> ** Testing ncvpt1 ...
> ** Testing ncvgt1 ...
> error in ncvgt1 byte:          0 .ne.        122
> error in ncvgt1 short:          0 .ne.         10
> error in ncvgt 1 float:   0.000000E+00 not close to       3.141590
> error in ncvgt1 double:  0.000000000000000E+000 not close to
>  3.333333333000000E-001
> ** Testing ncvpt ...
> ** Testing ncopn, ncinq, ncdinq, ncvinq, ncanam, ncainq ...
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> ** Testing ncvgt, ncvgtc ...
> ncvgt of bytes, got           0 .ne.          97
> ncvgt of bytes, got           0 .ne.          98
> ncvgt of bytes, got           0 .ne.          99
> ncvgt of bytes, got           0 .ne.         100
> ncvgt of short, got           0 .ne.          10
> ncvgt of short, got           0 .ne.          11
> ncvgt of short, got           0 .ne.          12
> ncvgt of short, got           0 .ne.          13
> float error in ncvgt
> ....(omitted 398 same error messages here!)....
> float error in ncvgt
> double error in ncvgt:          1   0.000000000000000E+000.ne.
> 1.000000
> ....(omitted 398 error messages whose value between (1,400) )....
> double error in ncvgt:        400   0.000000000000000E+000.ne.
> 400.000000
> ** Testing ncagt, ncagtc ...
> ** Testing ncredf, ncdren, ncvren, ncaren, ncendf ...
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> error in ncvinq or ncvdef
> ** Testing ncacpy ...
> ** Testing ncadel ...
> ** Testing fill values ...
> error in byte fill value
> error in double fill value
> error in float fill value
> error in short fill value
> =================end of message========================

Obviously we have more problems here than I thought.

> Is it the time dump "MS Power Station Fortran" off use
> the newer "Digital Fortran(version 5.0)" instead?

Well, since I don't have the older compilers, I'm not
going to be able to help you out much with this problem.
It is tricky getting all the preprocessor directives correct
for a give combination of Fortran and C compilation enviroments.
On the unix systems, the 'configure' script can set things
up pretty automatically. On Win32, VMS, and the like we
pretty much do it by hand.

If you do get the newer fortran compiler, you might
need the newer C compiler as well.

-glenn