CDF Version 2.3, MS-DOS Systems Common Data Format (CDF) is a conceptual data abstraction for storing multi-dimensional data sets. The basic component of CDF is a software programming interface that is a device independent view of the CDF data model. The application developer is insulated from the actual physical file format for reasons of conceptual simplicity, device independence, and future expandability. Version 2 of CDF has been rewritten to be portable across a wide variety of platforms. CDF files created on any given platform can be transported to any other platform on to which CDF is ported and used with any CDF tools or layered applications. A more detailed introduction to CDF can be found in the CDF User's Guide. For MS-DOS systems, the library and executables are available in addition to the source code. The source code is provided in case you are using a C compiler other than those currently supported (which affects whether or not you will be able to use the precompiled library). The currently supported C compilers are Microsoft C and Borland C. FORTRAN support is provided for Microsoft FORTRAN. The CDF library and toolkit programs are written in C. A C compiler (ANSI standard preferred) is necessary to build the CDF distribution (if you need to). The CDF distribution also contains several FORTRAN source files which support the FORTRAN interface. If a FORTRAN compiler is not available, the distribution can be built without FORTRAN support. Documentation for CDF consists of the CDF User's Guide. This document has been written using LaTeX and is available as an "Adobe" PostScript file named `cdfug.ps'. It should be printable on any PostScript printer capable of printing "Adobe" PostScript. If you have trouble printing the CDF User's Guide, contact us (we can mail you a hard copy if necessary). The document will be printed so that you can copy it to double-sided and then bind it like a book. Requests for assistance or documentation should be sent to: Internet -- CDFSUPPORT@NSSDCA.GSFC.NASA.GOV (128.183.36.23) If you get the distribution, please send CDFSUPPORT a message so we can put you on our mailing list. That way you will be notified of new releases. MS-DOS Installation Instructions -------------------------------- CDF V2.3 is distributed for MS-DOS systems in self-extracting LHA files (which are executables with extensions .EXE). The LHA utility is similar to ZIP/UNZIP. When a self-extracting LHA file is executed, one or more files and/or directories will be created in the current directory. You do not need the LHA utility to extract the CDF distribution. The CDF distribution is broken into several parts so that you can install only the parts that you need. To extract the files from a self-extracting LHA file, simply set your default directory to where the files (or directory tree of files) are to be placed and execute the appropriate executable as follows... > CDF23nnn where "nnn" is ETC, BIN, etc. (you may have to specify the full pathname of the self-extracting LHA file if it is not in the current directory). This will cause one or more files (and possibly new directories) to be created in your current directory (and possibly the parent directory of the current directory - see CDF23MS.EXE and CDF23BOR.EXE). You may install the CDF distribution in any directory you wish. It would probably be a good idea to create a new directory in which to install the CDF distribution. The contents of each self-extracting LHA file are as follows... CDF23ETC.EXE Documentation, online help files, etc. All CDF installations should extract the contents of this file. A directory named .\HELP will be created. CDF23BIN.EXE Toolkit executables (binaries). If you plan to install the CDF source code and compile/link the entire distribution, you don't need to extract the executables from this file. They will be created when you make the distribution. If you don't plan to extract the source code, extract the executables in this file. A directory named .\BIN will be created. CDF23PGM.EXE Library and include files used by applications. If you are going to extract the source code and make the distribution yourself, this file is not needed. If not, extract the library and include files from this file. Directories named .\LIB and .\INCLUDE will be created. The .\LIB directory will contain two libraries, LIBCDF.BOR and LIBCDF.MS (for Borland C and Microsoft C respectively). You should rename the appropriate one LIBCDF.LIB depending on the compiler you will be using (you may also want to delete the other one to save disk space). Each library has been built using the LARGE memory model (the library must be built with either the LARGE or HUGE memory model). If you require the HUGE memory model, rebuild the distribution following the steps below (the toolkit programs do not have to be rebuilt). CDF23SAM.EXE Sample CDFs. If extracted, a directory named .\SAMPLES will be created. CDF23UG.EXE The CDF User's Guide (an "Adobe" PostScript file). CDF23BOR.EXE Source code for use with Borland C. CDF23MS.EXE Source code for use with Microsoft C/FORTRAN. One of these is only needed if you are going to compile/link the distribution with your own compiler. If you are going to extract the source code, you must first create a directory named SRC and set your current directory to that directory. When CDF23BOR.EXE (or CDF23MS.EXE) is executed, the following directories will be created... INCLUDE LIB TEST TOOLS TOOLS\CURSES TOOLS\CURSES\SRC TOOLS\CURSES\INCLUDE TOOLS\CURSES\LIB ...will be created in the SRC directory. You should need set your current directory back up one level. ONLY EXTRACT ONE OF THESE! These self-extracting LHA files are distributed via anonymous FTP or on one or more floppy disks. Each disk should contain this README file in addition to one or more of the self-extracting LHA files. You can install the parts of the CDF distribution you need in any order you wish. Several of the self-extracting LHA files are too large to fit on floppy disks having limited capacity. If this is the case, the self-extracting LHA file will be split across two or more floppy disks with the SLICE/SPLICE utilities. Each floppy disk will be labeled appropriately (eg. `1 of 2' and `2 of 2'). To recombine the self-extracting LHA file (in your current directory), insert the floppy disk labeled `1 of...' and use the command > x:SPLICE x: where `x' is the appropriate floppy disk drive letter. The SPLICE utility will prompt for additional floppy disks (in numerical order) as they are needed. When SPLICE completes execution, the self-extracting LHA file may be used as if it was simply copied from the floppy disk(s). ******************************************************************************* * Skip the following section if you do not need to rebuild the entire CDF * * distribution. * ******************************************************************************* If you need to rebuild the CDF distribution, perform the following steps. All of these steps require your current directory to be set to the top-level CDF directory (the directory containing the SRC subdirectory). 1. Make files are provided with the source code (with extensions of .BOR and .MS for Borland C and Microsoft C/FORTRAN respectively). If you are using a different compiler, new make files must first be created (using the existing make files as templates). Then compiler/link the entire distribution using the "make" utility provided with your compiler. (If "make" does not exist, you will have to use an alternative method to compile/link the distribution.) Examples follow... make -fmakefile.bor -DBCLOC= all (Borland C) nmake /F makefile.ms all (Microsoft C) If a FORTRAN compiler is not available (when using Microsoft C), use `all.noF' instead of `all'. For Borland C, is the full pathname of the top-level directory of your Borland C compiler installation. If you are building the distribution with the HUGE memory model, you will first have to edit the make files in SRC\LIB, SRC\TEST, and SRC\TOOLS to indicate `H' rather than `L' for the macro MODEL. You can also build just a part of the distribution by setting your current directory to the appropriate source subdirectory (eg. SRC\LIB to only rebuild the CDF library). If you are using a compiler other than Borland C or Microsoft C/FORTRAN, you will also have to build a Curses library for your compiler. The Curses source is found in .\SRC\TOOLS\CURSES\SRC (in a self-extracting LHA file). A sample make file is also found in that directory. The Curses library you create should be put (overwritting the existing Curses library) in the directory named .\SRC\TOOLS\CURSES\LIB. 2. Test that the distribution has been built correctly with the following... make -fmakefile.bor test (Borland C) nmake /F makefile.ms test (Microsoft C) ...or whatever is appropriate for the compiler you are using. If a FORTRAN compiler is not available (when using Microsoft C), use `test.noF' instead of `test'. 3. Install the library, include files, and toolkit programs with the following... make -fmakefile.bor install (Borland C) nmake /F makefile.ms install (Microsoft C) ...or whatever is appropriate for the compiler you are using. This will cause the directories named .\LIB, .\INCLUDE, and .\BIN to be created. ******************************************************************************* * Continue reading here... * ******************************************************************************* Whether you built the distribution from the source code or simply extracted the library, include files, and toolkit executables, you should now have three directories containing the following files... .\INCLUDE CDF.H CDF.INC .\LIB LIBCDF.LIB .\BIN CDFBROW.EXE (CDFbrowse) CDFWALK.EXE (CDFwalk) CDFLIST.EXE (CDFlist) CDFSCOPE.EXE (CDFscope) CDFCVT.EXE (CDFconvert) CDFCMP.EXE (CDFcompare) CDFSTATS.EXE (CDFstats) CDFDUMP.EXE (CDFdump) -undocumented CDFEDIT.EXE (CDFedit) -undocumented CDFINQ.EXE (CDFinquire CDF2SKT.EXE (SkeletonTable) SKT2CDF.EXE (CDFskeleton) CDFDIR.BAT (CDFdir The name used in the CDF User's Guide for each toolkit program is shown in parenthesis. You may want to setup aliases for the toolkit programs (if you have that capability) to match the names used in the CDF User's Guide. You should set an environment variable named CDF_HELP to the full pathname of the .\HELP directory. That will allow the toolkit programs to locate their online help files.