20030311: HDF portion of v2002 build failure on Solaris 8 (cont.)

>From: "Anderson, Alan C. " <address@hidden>
>Organization: St. Cloud State
>Keywords: 200303112223.h2BMNIB2013810 McIDAS Sun Solaris HDF

Alan,

>The return from the cc -V  and f77 -V commands are
>
>cc -V
>cc:SunWorkshop 6 update 1 C 5.2  2000/09/11
>
>f77 -V
>f77: SunWorkshop 6 update Fortran77  5.2  2000/09/11

OK, these should work fine in building v2002.

>I need to check again but I think my problem is that I did not
>set the environment variables as directed in the netCDF sertion
>of the Notes and Warnings.   Somehow since I knew that I was 
>no longer using gcc, and that the Sun compilers were already there
>I left this out.  

This step is needed.  The settings are used in the netCDF, HDF,
and Tcl/Tk portions of the build.

>The directions now seem clear that regardless of which compiler
>one might be using,  we need to execute a set of  setenv  commands
>to identify things correctly.  

Right.

>I will do the clobber,  set my environment variables for the compilers
>and try it again.

Sounds good.

>What puzzles me is that the build went as far as it did.

The main McIDAS portion of the build figures out what it needs in the
build based on content in the 'mccomp' script (which gets made from the
mccomp.sh script in the distribution).  The netCDF, HDF, and Tcl/Tk
portions of the build use the configure approach that is common to
a large portion of Unix packages.  They all try to figure out what
compiler to use and what flags to use for those compilers.  Evidently,
and this is what I have been finding with the HDF portion of the build,
the HDF section's configure is not as "smart" as the netCDF or Tcl/Tk
portions.  In any case, if the environment variables like CC are set,
_all_ of the sections will use the settings in their builds.  So,
setting the variables makes the build environment predictable.

>Thanks

Please let me know how the build goes after you do the 'make clobber'
and set the environment variables.

Tom