|
|
I. Introduction.TRANSP is a collection of tools for analysis of tokamak data. A complete unix based TRANSP now exists. This is a complete system, with a version of Ufiles for specification of input data, and a copy of rplot for examination of output data. Standalone TRANSP on Unix is intended for use as a research tool, and as a basis for shared development and improvement of the code. It has the following characteristics: (1) The code, though originally developed under VAX/VMS, has now been under development on unix machines for over ten years. Most of the code is in fortran-77, but important sections are written in C and fortran-90, as well. Versions of TRANSP have run on DECstations, DEC Alpha (VMS and unix), Sparc-10s, Silicon Graphics, Linux, and parts of the TRANSP system have run on HP and IBM RS-6000 systems. All code, when compiled, goes through a cpp-based fortran preprocessing step. In the source one will find lines like: #ifdef __VMS
OPEN(UNIT=LUN,FILE='foo.data',STATUS='OLD',
> CARRIAGECONTROL='LIST')
#endif
#ifdef __UNIX
OPEN(UNIT=LUN,FILE='foo.data',STATUS='OLD')
#endif
which specify operating system specific code. (2) The unix TRANSP front end codes support ascii and several variant types of binary compressed Ufiles. The more modern compressed format is portable between machines of different types (including vms <-> unix) and involves no loss of precision for IEEE single precision floating point ddata. (The script `ufcheck' can be used to identify the format of Ufiles in a directory; the script `ufcompress' can be used to convert all Ufiles in a directory and its subdirectories into portable compressed binary). NetCDF and HDF Ufiles format variants are available as well -- provided these libraries are available on the system and included in the TRANSP link. For more details see $CODESYSDIR/source/doc/netcdf.doc; it is not necessary to bother with this on a first installation of TRANSP. (3) The TRANSP unix environment is based on the C shell (csh). The reader of this document is assumed to have a working knowledge of unix and C shell. (4) The code documentation (in $CODESYSDIR/doc and $CODESYSDIR/source/doc) is a straight copy of VAX documentation. However, some TRANSP documentation, in particular the namelist help file, has been put up on the web (available under http://w3.pppl.gov/transp). (5) The code includes interactive graphics facilities based on the SG ("Scientific Graphics") library (Marilee Thompson, PPPL -- now maintained by Tina Ludescher). The codes have been tested on Macintosh computers running Versaterm and on X terminals running xterm. An X windows based sglib output viewer, xtc, is also available (see the NTCC modules library, http://w3.pppl.gov/NTCC). Standalone unix TRANSP lacks a number of features that might need to be added for a "production environment":
The PPPL production system remains VAX-based and relies on proprietary software and hardware, so it is not exportable at this time. NOTE (Jan 1999): experimental software for networked queue management and automatic recovery after system crashes is now available. For more information see $CODESYSDIR/source/doc/qmanager.doc. ............. This document concerns itself primarily with the installation of unix TRANSP. Instructions for operation and/or modification of the code are given elsewhere. There are two major phases to the installation process: (1) obtaining a copy of the code sources, and (2) building and testing the executable binaries. II. Obtaining a copy of TRANSP
PREREQUISITE HARDWARE and SOFTWARE:See `remote_site.doc' for a detailed list of requirements -- available by anonymous ftp in the ftp.pppl.gov:pub/transp/codata/doc directory. Also in $CODESYSDIR/source/doc/remote_site.doc. The instructions for obtaining a copy of the code are now in the file available by anonymous ftp at ftp.pppl.gov:pub/transp/codata. The code itself is NOT available by anonymous ftp. See the README file for details. NOTE -- 29 Sep 1993 -- a method has been developed to check out
subsets of TRANSP (i.e. NOT the whole code, but enough to be able to work on a subset of
the code). Such a check-out requires much less disk space. If you need this, follow the
instructions in the README file. |