Changes from Version edited by Vince: Created Makefile. unpack.c : No changes. Looks good. Thats was a good idea to provide this. Needs a little documentation maybe. renamed subroutine in freadpathP.f to the same name. renamed sample_readpathP.f sample_readpathP_f.f renamed sample_readpathP.c sample_readpathP_c.f This makes it easier to create separate targets in the Makefile. sample_readpathP_f.f: Defined RSIZE variable to indicated record length. Users have to edit this variable depending on the machine they are using. On SUN's and HP's we need to specify the record length for unformatted IO to the recordlength in bytes. Other machines (DEC, SGI I believe) want the record length in longwords (4 bytes). This needs to be documented somewhere in Release Notes. changed hard-coded input file name to sample.hdf. sample_readpathP_c.f: Maybe we don't even need this anymore. But maybe yes. While the unpack function is something users should be able to use without editing, the sample_readpathP_X programs are truely only meant to be examples. Unpack and sample_readpath_x may diverge at some point (What do you think ? ) Things to do: The unpack function should really check if the *.exhdf files are in the current working directory and report an error if they aren't. That's an easy trap to fall into, so I think that would be a good idea. Vince can you add this? Axel 960212 Axel 960912: sample_readpathP_f.f would blow up under Solaris 2.5. Fixed call to sfgainfo (not really needed) : ret = sfgainfo(hdf_id,index,attr_name, number_type, & length) Since the variable attribute_name is passed as a fixed string and sfgainfo expects returns data using attribute_name instead of attr_name doesn't work. Created new variable character*64 in which attr_name is returned. Axel: 960913 Fixed Makefile to work for HP, SunOS and Solaris versions. Axel: 960925 Changed readpathP.pro to resolve functions included in this routine. It is necessary to .run readpathP to get it working still or else break the program into subprograms.