950814: Problems with NETCDF-232pl2

Mike,

> Subject: Problems with NETCDF-232pl2
> To:       address@hidden
> From: Mike Mikitiuk <address@hidden>
> Organization: University of Western Ontario
> Keywords: 199508141916.AA06907

In the above message you wrote:

> I am a GMT3.0 user and I am having difficulties getting NETCDF-232pl2 to
> work on an IBM RS6000 Model 3AT running AIX 3.2.

> 
> I ran 'configure' using 
> 
>    configure --prefix=/pcdisk/gmt3/netcdf-232pl2
> 
> The 'Makefile' is included at the end of this message.
> 
> I then ran
> 
>    make all
> 
> and 
> 
>    make test
> 
> The 'script' of these 'makes' is also included at the end of this message.
> The 'test' make bombs out in 'cdftest.c', line 393.
> Can you give me a 'fix' to this problem ?

This problem is caused by a bug in the AIX 3.2 C compiler when it
compiles the file `libsrc/string.c' with optimization enabled.  To
workaround this bug, do the following:

    1.  Enter the subdirectory `libsrc/'.

    2.  Remove the file `string.o' so that it will be recreated.

    3.  Add the following rule to the makefile template file 
        `Makefile.in' somewhere after the `all' target but before 
        the `include ../port/master.mk':

        # The following rule exists as a workaround for a bug in the AIX 3.2 C
        # compiler (it doesn't optimize correctly).
        #
        string.o:       string.c
                case `uname -sv` in \
                    'AIX 3') \
                        CFLAGS=`echo $(CFLAGS) | sed 's/-O[23]//g;s/-O//g'`; \
                        $(CC) -c $$CFLAGS $(CPPFLAGS) string.c; \
                        ;; \
                    *) \
                        $(CC) -c $(CFLAGS) $(CPPFLAGS) string.c; \
                        ;; \
                esac

    4.  Go up one directory to the top-level source directory.

    5.  Execute the script `./config.status' to remake the makefile 
        `libsrc/Makefile'.

    6.  Execute the command `make all'.

    7.  Execute the command `make test'.

The next release of the netCDF package won't have this problem.

--------
Steve Emmerson   <address@hidden>