.RP
.de XS
.DS
.ps -1
.vs -2p
.ft CB
..
.de XE
.DE
.ft R
.ps
.vs
..
.TL
UNIX/IRAF Site Manager's Guide
.AU
Doug Tody
.AI
IRAF Group
.br
.K2 "" "" "\(dg"
.br
June 1989
.br
Revised September 1992

.AB
An IRAF \fIsite manager\fR is anyone who is responsible for installing and
maintaining IRAF at a site.  This document describes a variety of site
management activities, including configuring the device and environment
tables to provide reasonable defaults for the local site, adding interfaces
for new devices, configuring and using IRAF networking, the installation
and maintenance of layered software products (external packages),
and configuring a custom site LOCAL package so that local software may be
added to the system.  Background information on multiple architecture
support, shared library support, and the software management tools provided
with the system is presented.  The procedures for rebooting IRAF and
performing a sysgen are described.  The host system resources
required to run IRAF are discussed.
.AE

.pn 1
.bp
.ce
.ps +2
\fBContents\fR
.ps -2
.sp 3
.sp
1.\h'|0.4i'\fBIntroduction\fP\l'|5.6i.'\0\01
.sp
2.\h'|0.4i'\fBSystem Setup\fP\l'|5.6i.'\0\02
.br
\h'|0.4i'2.1.\h'|0.9i'Installing the System\l'|5.6i.'\0\02
.br
\h'|0.4i'2.2.\h'|0.9i'Configuring the Device and Environment Tables\l'|5.6i.'\0\02
.br
\h'|0.9i'2.2.1.\h'|1.5i'Environment definitions\l'|5.6i.'\0\02
.br
\h'|0.9i'2.2.2.\h'|1.5i'The template LOGIN.CL\l'|5.6i.'\0\03
.br
\h'|0.9i'2.2.3.\h'|1.5i'The TAPECAP file\l'|5.6i.'\0\03
.br
\h'|0.9i'2.2.4.\h'|1.5i'The DEVICES.HLP file\l'|5.6i.'\0\03
.br
\h'|0.9i'2.2.5.\h'|1.5i'The TERMCAP file\l'|5.6i.'\0\04
.br
\h'|0.9i'2.2.6.\h'|1.5i'The GRAPHCAP file\l'|5.6i.'\0\04
.br
\h'|0.9i'2.2.7.\h'|1.5i'Configuring IRAF networking\l'|5.6i.'\0\04
.br
\h'|0.9i'2.2.8.\h'|1.5i'Configuring the IRAF account\l'|5.6i.'\0\06
.br
\h'|0.9i'2.2.9.\h'|1.5i'Configuring user accounts for IRAF\l'|5.6i.'\0\06
.br
\h'|0.4i'2.3.\h'|0.9i'Tuning Considerations\l'|5.6i.'\0\06
.br
\h'|0.9i'2.3.2.\h'|1.5i'Stripping the system to reduce disk usage\l'|5.6i.'\0\06
.sp
3.\h'|0.4i'\fBSoftware Management\fP\l'|5.6i.'\0\07
.br
\h'|0.4i'3.1.\h'|0.9i'Multiple architecture support\l'|5.6i.'\0\07
.br
\h'|0.4i'3.2.\h'|0.9i'Shared libraries\l'|5.6i.'\0\08
.br
\h'|0.4i'3.3.\h'|0.9i'Layered software support\l'|5.6i.'\0\09
.br
\h'|0.4i'3.4.\h'|0.9i'Software management tools\l'|5.6i.'\0\10
.br
\h'|0.4i'3.5.\h'|0.9i'Modifying and updating a package\l'|5.6i.'\0\11
.br
\h'|0.4i'3.6.\h'|0.9i'Installing and maintaining layered software\l'|5.6i.'\0\12
.br
\h'|0.4i'3.7.\h'|0.9i'Configuring a custom LOCAL package\l'|5.6i.'\0\13
.br
\h'|0.4i'3.8.\h'|0.9i'Updating the full IRAF system\l'|5.6i.'\0\13
.br
\h'|0.9i'3.8.1.\h'|1.5i'The BOOTSTRAP\l'|5.6i.'\0\14
.br
\h'|0.9i'3.8.2.\h'|1.5i'The SYSGEN\l'|5.6i.'\0\14
.br
\h'|0.9i'3.8.3.\h'|1.5i'Localized software changes\l'|5.6i.'\0\15
.sp
4.\h'|0.4i'\fBGraphics and Image Display\fP\l'|5.6i.'\0\17
.br
\h'|0.4i'4.1.\h'|0.9i'Using the workstation with a remote compute server\l'|5.6i.'\0\17
.sp
5.\h'|0.4i'\fBInterfacing New Graphics Devices\fP\l'|5.6i.'\0\17
.br
\h'|0.4i'5.1.\h'|0.9i'Graphics terminals\l'|5.6i.'\0\17
.br
\h'|0.4i'5.2.\h'|0.9i'Graphics plotters\l'|5.6i.'\0\17
.br
\h'|0.4i'5.3.\h'|0.9i'Image display devices\l'|5.6i.'\0\18
.sp
6.\h'|0.4i'\fBHost System Requirements\fP\l'|5.6i.'\0\18
.br
\h'|0.4i'6.1.\h'|0.9i'Memory requirements\l'|5.6i.'\0\19
.br
\h'|0.4i'6.2.\h'|0.9i'Disk requirements\l'|5.6i.'\0\19
.sp
\fBAppendix A.\0The IRAF Directory Structure\fP\l'|5.6i.'\0\19
.nr PN 0
.bp

.NH
Introduction
.PP
Once the IRAF system has been installed it will run, but there remain many
things one might want to do to tailor the system to the local site.
Examples of the kinds of customizations one might want to make are the
following.
.RS
.IP \(bu
Edit the default IRAF environment definitions to provide reasonable
defaults for your site.
.IP \(bu
Make entries in the device descriptor tables for the devices in use at
your site.
.IP \(bu
Code and install new device interfaces.
.IP \(bu
Enable and configure IRAF networking, e.g., to permit remote image
display, tape drive, or file access.
.IP \(bu
Perform various optimizations, e.g., stripping the system to reduce disk
usage.
.IP \(bu
Extend the system by installing layered software products.
.IP \(bu
Configure a custom LOCAL package so that locally developed software
may be installed in the system.
.RE
.PP
This document provides sufficient background information and instructions to
guide the IRAF site manager in performing such customizations.  Additional
help is available via the IRAF HOTLINE (602 323-4160), or by sending mail to
\f(CWiraf@noao.edu\fR (internet) or \f(CW5355::iraf\fP (SPAN).
Contributions of interfaces developed for new devices, or any other software
of general interest, are always welcome.
.PP
The IRAF software is organized in a way which attempts to isolate, so far as
possible, the files or directories which must be modified to tailor the
system for the local site.  Most or all changes should affect only files in
the local, dev, and hlib (unix/hlib) directories.  Layered software
products, including locally added software, reside outside of the IRAF core
system directory tree and are maintained independently of the core system.
.PP
A summary of all modifications made to the IRAF system for a given IRAF
release is given in the \fIRevisions Summary\fR distributed with the
system.  Additional information will be found in the system notes files
(notes.v29, notes.v210, etc.) in the iraf/local and iraf/doc directories.
This is the primary source of technical documentation for each release and
should be consulted if questions arise regarding any of the system level
features added in a new release of the core system.

.bp
.NH
System Setup
.NH 2
Installing the System
.PP
The procedure for installing or updating a UNIX/IRAF system is documented in
the \fIIRAF Installation Guide\fR distributed with the system.  A custom
installation guide is provided for each platform on which IRAF is supported.
.PP
In short, an IRAF tape or network distribution is obtained and installed
according to the instructions.  The result is a full IRAF system, including
both sources and executable binaries for the architectures to be supported.
The system will have been modified to reflect the new IRAF root directory
and should run, but will otherwise be a generic IRAF distribution.  To get
the most out of an IRAF installation it will be necessary to perform some of
the additional steps outlined in the remainder of this document.

.NH 2
Configuring the Device and Environment Tables
.PP
Teaching IRAF about the devices, network nodes, external programs, and other
special resources available at a site is largely a matter of editing a
standard set of device descriptor and environment setup files, all of which
are simple text files.  The versions of these files provided with the
distribution are simply those in use on the NOAO system from which the tapes
were made, at the time the tapes were generated.  Hence while these files
may be useful as examples of properly configured descriptor files, the
defaults, and many specific device entries, will in many cases be
meaningless for a different site.  This is harmless but it may be confusing
to the user if, for example, the default printer doesn't exist at your
site.
.PP
The device and environment files also contain much material which any site
will need, so care must be taken when editing the files.  Important changes
may be made to the global portions of these files as part of any IRAF
release.  To facilitate future updates, it is wise where possible to isolate
any local changes or additions so that they may simply be extracted and
copied into the new (distributed) version of the file in a future update.
.NH 3
Environment definitions
.PP
Since IRAF is a machine and operating system independent, distributed system
it has its own environment facility apart from that of the host system.
Host system environment variables may be accessed as if they are part of the
IRAF environment (which is sometimes useful but which can also be
dangerous), but if the same variable is defined in the IRAF environment it
is the IRAF variable which will be used.  The IRAF environment definitions,
as defined at CL startup time, are defined in a number of files in the
unix/hlib directory.  Chief among these is the \fBzzsetenv.def\fR file.
Additional user modifiable definitions may be given in the template
\fBlogin.cl\fR file (see \(sc2.2.2).
.PP
The zzsetenv.def file contains a number of environment definitions.
Many of these define IRAF logical directories and should be left alone.
Only those definitions in the header area of the file should need to be
edited to customize the file for a site.  It is the default editor,
default device, etc. definitions in this file which are most likely to
require modification for a site.
.PP
If the name of a default device is modified, the named device must also have
an entry in the \fBtermcap\fR file (terminals and printers) or the
\fBgraphcap\fR file (graphics terminals and image displays) in iraf/dev.
There must also be an \fIeditor\f(CW.ed\fR file in dev for the
default editor; \fIedt\fR, \fIemacs\fR, and \fIvi\fR are examples of
currently supported editors.
.PP
Sample values of those variables most likely to require modification for
a site are shown below.
.XS
set editor      = "vi"
set printer     = "lpr"
set stdplot     = "lpr"
set stdimage    = "imt512"
.XE
.PP
For example, you may wish to change the default editor to "emacs", the
default printer to "lw5", or the default image display to "imt800".  Note
that the values of terminal and stdgraph, which also appear in the
zzsetenv.def file, have little meaning except for debugging processes run
standalone, as the values of the environment variables are reset
automatically by \fIstty\fR at login time.  The issues of interfacing new
graphics and image display devices are discussed further in \(sc5.
.NH 3
The template LOGIN.CL
.PP
The template login.cl file hlib$login.cl, is the file used by \fImkiraf\fR
to produce the user login.cl file.  The user login.cl file, after having
possibly been edited by the user, is read by the CL every time a new CL is
started, with the CL processing all environment and task definitions,
package loads, etc., in the login file.  Hence this file plays an important
role in establishing the IRAF environment seen by the user.
.PP
Examples of things one might want to change in the template login.cl
are the commented out environment definitions, the commented out CL
parameter assignments, the foreign task definitions making up the default
\f(CWuser\fR package, and the list of packages to be loaded at startup
time.  For example, if there are host tasks or local packages which
should be part of the default IRAF operating environment at your site,
the template login.cl is the place to make the necessary changes.
.NH 3
The TAPECAP file
.PP
Beginning with V2.10 IRAF magtape devices are described by the tapecap file,
dev$tapecap.  This replaces the "devices" file used in earlier versions of
IRAF.  The tapecap file describes each local magtape device and controls all
i/o to the device, as well as device allocation.
.PP
The tapecap file included in the distributed system includes some generic
device entries such as "mtxb1" (Exabyte unit 1), "mtwd0" (WangDAT unit 0),
and so on which you may be able to use as-is to access your local magtape
devices.  The actual list of generic device entries provided is system
dependent, so consult the tapecap file in your installed system for a list
of the currently interfaced devices.  Most likely you will want to add some
device aliases, and you may need to prepare custom device entries for local
devices.  There must be an entry in the tapecap file for a magtape device in
order to be able to access the device from within IRAF.
.PP
Instructions for adding devices to the tapecap file are given in the
document \fIIRAF Version 2.10 Revisions Summary\fR, in the discussion of
the new magtape system.
.NH 3
The DEVICES.HLP file
.PP
All physical devices that the user might need to access by name should be
documented in the file dev$devices.hlp.  Typing
.XS
cl> help devices
.XE
or just
.XS
cl> devices
.XE
in the CL will format and output the contents of this file.  It is the IRAF
name of the device, as given in files such as termcap, graphcap, and
tapecap, which should appear in this help file.
.NH 3
The TERMCAP file
.PP
There must be entries in this file for all local terminal and printer
devices you wish to access from IRAF (there is currently no "printcap" file
in IRAF).  The entry for a printer contains one special device-specific
entry, called \f(CWDD\fR.  This consists of three fields: the device name,
e.g. "node!device", the template for the temporary spoolfile, and the UNIX
command to be used to dispose of the file to the printer.  On most UNIX
systems it is not necessary to make use of the node name and IRAF networking
to access a remote device since UNIX \fIlpr\fR already provides this
capability, however it might still be useful if the desired device does not
have a local \fIlpr\fR entry for some reason.
.PP
If you have a local terminal which has no entry in the IRAF termcap file,
you probably already have an entry in the UNIX termcap file.  Simply copy it
into the IRAF file; both systems use the same termcap database format and
terminal device capabilities.  However, if the terminal in question is a
graphics terminal with a device entry in the graphcap file, you should
add a `\f(CW:gd\fR' capability to the termcap entry.  If the graphcap entry
has a different name from the termcap entry, make it `\f(CW:gd=\fIgname\fR'.
.NH 3
The GRAPHCAP file
.PP
There must be entries in the graphcap file for all graphics terminals, batch
plotters, and image displays accessed by IRAF programs.  New graphics
terminals will need a new entry.  The IRAF file gio$doc/gio.hlp
contains documentation describing how to prepare graphcap device entries.  A
printed copy of this document is available from the iraf/docs directory in
the IRAF network archive.  However, once IRAF is up you may find it easier
to generate your own copy using \fIhelp\fR, as follows:
.XS
cl> help gio$doc/gio.hlp fi+ | lprint
.XE
which will print the document on the default IRAF printer device (use the
"device=" hidden parameter to specify a different device).  Alternatively,
to view the file on the terminal,
.XS
cl> phelp gio$doc/gio.hlp fi+
.XE
.PP
The help pages for the IRAF tasks \fIshowcap\fR and \fIstty\fR should also
be reviewed as these utilities are useful for generating new graphcap
entries.  The i/o logging feature of \fIstty\fR is useful for determining
exactly what characters your graphcap device entry is generating.  The
\fIgdevices\fR task is useful for printing summary information about the
available graphics devices.
.PP
Help preparing new graphcap device entries is available if needed.  We ask
that new graphcap entries be sent back to us so that we may include them in
the master graphcap file for all to benefit.
.NH 3
Configuring IRAF networking
.PP
The dev directory contains several files (\f(CWhosts\fR,
\f(CWirafhosts\fR, and \f(CWuhosts\fR) used by the IRAF network interface.
IRAF networking is used to access remote image displays, printers, magtape
devices, files, images, etc. via the network.  Nodes do not necessarily have
to have the same architecture, or even run the same operating system, so
long as they can run IRAF.
.PP
To enable IRAF networking for a UNIX/IRAF system, all that is necessary is to
edit the "hosts" file.  Make an entry for each logical node, in the format
.XS
\fInodename\fR [ \fIaliases\fR ] ":" \fIirafks.e-pathname\fR
.XE
following the examples given in the hosts file supplied with the
distribution (which is the NOAO/Tucson hosts file).  Note that there may be
multiple logical entries for a single physical node.
.PP
The "uhosts" file is not used by UNIX/IRAF systems hence does not need to
be modified (it used by VMS/IRAF).  The "irafhosts" file is the template
file used to create user .irafhosts files.  It does not have to be modified,
although you can do so if you wish to change the default parameter values
given in the file.
.PP
To enable IRAF networking on a particular IRAF host, the \fBhostname\fR for
the host machine must appear as a primary name or alias somewhere in the
IRAF hosts table.  During process startup, the IRAF VOS looks for the system
name for the current host and automatically disables networking if this name
is not found.  Hence IRAF networking is automatically disabled when the
distributed system is first installed - unless you are unlucky enough to
have installed the system on a host with the same name as one of the nodes
in the NOAO host table.
.PP
Once IRAF networking is configured, the following command may be typed in
the CL to verify that all is well:
.XS
cl> netstatus
.XE
This will print the host table and state the name of the local host.
Read the output carefully to see if any problems are reported.
.PP
For IRAF networking to be of any use, it is necessary that IRAF be installed
on at least two systems.  In that case either system can serve as the server
for an IRAF client (IRAF program) running on the other node.  It is not
necessary to have a separate copy of IRAF on each node, i.e., a single copy
of IRAF may be NFS mounted on all nodes (you will need to run the IRAF
\fIinstall\fR script on each client node).  If it is not possible to install
IRAF on a node for some reason (either directly or using NFS) it is possible
to manage by installing only enough of IRAF to run the IRAF kernel server.
Contact IRAF site support if you need to configure things in this manner.
.PP
UNIX IRAF systems currently support only TCP/IP based networking.
Networking between any heterogeneous collection of systems is possible
provided they support TCP/IP based networking (virtually all UNIX-based
systems do).  The situation with networking between UNIX and VMS systems is
more complex.  V2.9 and earlier versions of VMS/IRAF support client-side
only TCP/IP using the third party Wollongong software.  For V2.10 we plan to
drop support for the Wollongong software and switch to the more
fully-featured Multinet instead (another third party product).  Contact the
IRAF project for further information on networking between UNIX and VMS
systems.
.PP
Once IRAF networking is enabled, objects resident on the server node may be
accessed from within IRAF merely by specifying the node name in the object
name, with a "\fInode!\fR" prefix.  For example, if \fIfoo\fR is a network
node,
.XS
cl> page foo!hlib$motd
cl> allocate foo!mta
cl> devstatus foo!mta
.XE
.PP
In a network of "trusted hosts" the network connection will be made
automatically, without a password prompt.  A password prompt will be
generated if the user does not have permission to access the remote node
with UNIX commands such as \fIrsh\fR.  Each user has a .irafhosts file in
their UNIX login directory which can be used to exercise more control over
how the system connect to remote hosts.  See the discussion of IRAF
networking in the \fIIRAF Version 2.10 Revisions Summary\fR, or in the V2.10
system notes file, for a more in-depth discussion of how IRAF networking
works.
.PP
To keep track of where files are in a distributed file system, IRAF uses
\fBnetwork pathnames\fR.  A network pathname is a name such as
"foo!/tmp3/images/m51.pix", i.e., a host or IRAF filename with the node name
prepended.  The network pathname allows an IRAF process running on any node
to access an object regardless of where it is located on the network.
.PP
Inefficiencies can result when image pixel files are stored on disks which
are cross-mounted using NFS.  The typical problem arises when imdir (the
pixel file storage directory) is set to a path such as "/data/iraf/user/",
where /data is a NFS mounted directory.  Since NFS is transparent to
applications like IRAF, IRAF thinks that /data is a local disk and the
network pathame for a pixel file will be something like "foo!/data/iraf"
where "foo" is the hostname of the machine on which the file is written.  If
the image is then accessed from a different network node the image data will
be accessed via an IRAF networking connection to node "foo", followed by an
NFS connection to the node on which the disk is physically mounted, causing
the data to traverse the network twice, slowing access and unnecessarily
loading the network.
.LP
A simple way to avoid this sort of problem is to include the server name
in the imdir, e.g.,
.XS
cl> set imdir = "server!/data/iraf/user/"
.XE
This also has the advantage of avoiding NFS for pixel file access - NFS is
fine for small files but can load the server excessively when used to access
bulk image data.
.PP
Alternatively, one can set imdir to a value such as "HDR$pixels/", or
disable IRAF networking for disk file access.  In both cases NFS will be
used for image file access.
.NH 3
Configuring the IRAF account
.PP
The IRAF account, i.e., what one gets when one logs into UNIX as "iraf",
is the account used by the IRAF site manager to work on the IRAF system.
Anyone who uses this account is in effect a site manager, since they have
permission to modify, delete, or rebuild any part of IRAF.  For these and
other reasons (e.g., concurrency problems) it is recommended that all routine
use of IRAF be performed from other accounts (user accounts).
.PP
If the system has been installed according to the instructions given in the
installation guide the login directory for the IRAF account will be
iraf/local.  This directory contains both a \f(CW.login\fR file
defining the environment for the IRAF account, and a number of other "dot"
files used to setup the IRAF system manager's working environment.
.PP
Most site managers will probably want to customize these files according to
their personal preferences.  In doing this please use caution to avoid losing
environment definitions, etc., which are essential to the correct operation
of IRAF, including IRAF software development.
.PP
The default login.cl file supplied in the IRAF login directory uses machine
independent pathnames and should work as-is (no need to do a \fImkiraf\fR -
in fact \fImkiraf\fR has safeguards against inadvertent use within the IRAF
directories and may not work in iraf/local).  It may be necessary to edit
the .login file to modify the way the environment variable \f(CWIRAFARCH\fR
is defined.  This variable, required for software development but optional
for merely using IRAF, must be set to the name of the desired machine
architecture, e.g., sparc, vax, rs6000, ddec, etc.  If it is set to the name
of an architecture for which there are no binaries, e.g., generic, the CL
may not run, so be careful.  The alias \fIsetarch\fR, defined in the iraf
account .login, is convenient for setting the desired architecture for IRAF
execution and software development.
.NH 3
Configuring user accounts for IRAF
.PP
User accounts should be loosely modeled after the IRAF account.  All that is
required for a user to run IRAF is that they run \fImkiraf\fR in their
desired IRAF login directory before starting up the CL.  Defining
\f(CWIRAFARCH\fR in the user environment is not required unless the user
will be doing any IRAF based software development (including IMFORT).
Programmers doing IRAF software development may wish to source
hlib$irafuser.csh in their .login file as well.

.NH 2
Tuning Considerations
.NH 3
Stripping the system to reduce disk usage
.PP
If the system is to be installed on multiple CPUs, or if a production
version is to be installed on a workstation, it may be necessary or
desirable to strip the system of all non-runtime files to save disk space.
This equates to deleting all the sources and all the reference manuals and
other documentation, excluding the online manual pages.  A special utility
called \fIrmfiles\fR (in the SOFTOOLS package) is provided for this
purpose.  It is not necessary to run \fIrmfiles\fR directly to strip the
system.  The preferred technique is to use "mkpkg strip" as in the following
example (this may be executed from either the host system or from within
IRAF).
.XS
% cd $iraf
% mkpkg strip
.XE
.PP
This will preserve all runtime files, permitting use of the standard system
as well as user software development.  Note that only the IRAF core system
is stripped, i.e., if you want to strip any external layered software
products, such as the NOAO package, a \fImkpkg strip\fR must be executed
separately for each - \fIcd\fR to the root directory of the external package
first.  A tape backup of a system should always be made before the system is
stripped; keep the backup indefinitely as it may be necessary to restore the
sources in order to, e.g., install a bug fix or add-on software product.

.NH
Software Management
.NH 2
Multiple architecture support
.PP
Often the computing facilities at a site consist of a heterogeneous network
of workstations and servers.  These machines will often have quite different
architectures.  Considering only a single vendor like Sun, as of 1992 one
sees the three major architectures SPARC, Motorola 68020, and Intel 80386,
and several minor variations on these architectures, i.e., the floating
point options for the Sun-3, namely the Motorola 68881 coprocessor, the Sun
floating point accelerator (FPA), and software floating point (Sun is trying
to phase some of these out but the need for multiple architecture support is
not likely to go away).  On the Decstation we currently support two
architectures, one (ddec) using the DEC Fortran compiler, and the other
(dmip) using the MIPS Risc Fortran compiler.  Other systems such as SGI/IRAF
or the VAXstation support only a single architecture.
.PP
Since IRAF is a large system it is undesirable to have to maintain a separate
copy of IRAF for each machine architecture on a network.  For this reason
IRAF provides support for multiple architectures within a single copy of IRAF.
To be accessible by multiple network clients, this central IRAF system will
typically be NFS mounted on each client.
.PP
Multiple architecture support is implemented by separating the IRAF sources
and binaries into different directory trees.  The sources are architecture
independent and hence sharable by machines of any architecture.  All of the
architecture dependence is concentrated into the binaries, which are collected
together into the so-called BIN directories, one for each architecture.
The BIN directory contains all the object files, object libraries, executables,
and shared library images for an architecture, supporting both IRAF execution
and software development for that architecture.  A given system can support
any number of BIN directories, and therefore any number of architectures.
.PP
In IRAF terminology, when we refer to an "architecture" what we really
mean is a type of BIN.  The correspondence between BINs and hardware
architectures is not necessarily one-to-one, i.e., multiple BINs can exist
for a single compiler architecture by compiling the system with different
compilation flags, as different versions of the software, and so on.
Examples of some currently supported software architectures are shown below.
.DS
.TS
center;
ci ci ci
l l l.
Architecture	System	Description
.sp
generic	any	no binaries (default IRAF configuration)
sparc	Sun-4	Sun SPARC (RISC) architecture, integral fpu
f68881	Sun-3	mc68020, 68881 floating point coprocessor
pg	Sun-4	Sun/IRAF compiled for profiling
ddec	Decstation	DEC Fortran version of DSUX/IRAF
dmip	Decstation	MIPS Risc Fortran version of DSUX/IRAF
rs6000	IBM	IBM RS/6000 running AIX
irix	SGI	SGI IRIX, MIPS cpu
f2c	Macintosh	A/UX, using Fortran-to-C translation and GCC
.TE
.DE
.PP
Most of these correspond to hardware architectures or floating point hardware
options.  The exceptions are the generic architecture, which is what
the distributed system is configured to by default (to avoid having any
architecture dependent binary files mingled with the sources), and the
"pg" architecture, which is not normally distributed to user sites,
but is a good example of a custom software architecture used for software
development.
.PP
When running IRAF on a system configured for multiple architectures,
selection of the BIN (architecture) to be used is controlled by the UNIX
environment variable \f(CWIRAFARCH\fR, e.g.,
.XS
% setenv IRAFARCH ddec
.XE
would cause IRAF to run using the ddec architecture, corresponding to the
BIN directory bin.ddec.  Once inside the CL one can check the current
architecture by entering one of the following commands (the output in each
case is shown as well).
.XS
cl> show IRAFARCH
ddec
.XE
or
.XS
.cc #
cl> show arch
.ddec
#cc
.XE
.LP
If IRAFARCH is undefined at CL startup time a default architecture will be
selected based on the current machine architecture, the available floating
point hardware, and the available BINs.  The IRAFARCH variable controls not
only the architecture of the executables used to run IRAF, but the libraries
used to link IRAF programs, when doing software development from within the
IRAF or host environment.
.PP
Additional information on multiple architecture support is provided in the
system notes file for V2.8, file doc$notes.v28.

.NH 2
Shared libraries
.PP
Among the UNIX based versions of IRAF, currently only Sun/IRAF supports
shared libraries, although we are looking into adding shared library support
to the other, mostly SysV based versions of IRAF.  SunOS has an unusually
powerful virtual file system architecture, and several years ago was one of
the few UNIX systems supporting shared, mapped access to files.  This is no
longer the case however, and nowadays most versions of UNIX provide some
sort of shared library facility.  Shared libraries result in a considerable
savings in disk space, so eventually we will probably implement the facility
for additional platforms.  In the meanwhile, if you are running IRAF on a
system other than a Sun this section can be skipped.
.PP
Sun/IRAF provides a shared library facility for SunOS 4.0 and later versions
of SunOS (but not for SunOS-3).  All architectures are supported.
So long as everything is working properly, the existence and use of the shared
library should be transparent to the user and to the site manager.
This section gives an overview of the shared library facility to point
the reader in the right direction in case questions should arise.
.PP
What the shared library facility does is take most of the IRAF system
software (currently the contents of the \f(CWex\fR, \f(CWsys\fR,
\f(CWvops\fR, and \f(CWos\fR libraries) and link it together into a special
sharable image, the file \f(CWS\fIn\fP.e\fR in each core system BIN
directory (\fIn\fR is the shared image version number, e.g. "S6.e").  This
file is mapped into the virtual memory of each IRAF process at process
startup time.  Since the shared image is shared by all IRAF processes, each
process uses less physical memory, and the process pagein time is reduced,
speeding process execution.  Likewise, since the subroutines forming the
shared image are no longer linked into each individual process executable,
substantial disk space is saved for the BIN directories.  Link time is
correspondingly reduced, speeding software development.
.PP
The shared library facility consists of the \fBshared image\fR itself,
which is an actual executable image (though not runnable on all systems),
and the \fBshared library\fR, contained in the library lib$libshare.a,
which defines each VOS symbol (subroutine), and which is what is linked
into each IRAF program.  The shared library object module does not consume
any space in the applications program, rather it consists entirely of symbols
pointing to \fBtransfer vector\fR slots in the header area of the shared
image.  The transfer vector slots point to the actual subroutines.
.PP
When an IRAF program is linked with \fIxc\fR, one has the option of linking
with either the shared library or the individual system libraries.  Linking
with the shared library is the default; the \f(CW-z\fR flag disables linking
with the shared library.  In the final stages of linking \fIxc\fR runs the
HSI utility \fIedsym\fR to edit the symbol table of the output executable,
modifying the shared library (VOS) symbols to point directly into the shared
image (to facilitate symbolic debugging), optionally deleting all shared
library symbols, or performing some other operation upon the shared library
symbols, depending upon the \fIxc\fR link flags given.
.PP
At process startup time, upon entry to the process main (a C main for
Sun/IRAF) the shared image will not yet have been mapped into the address
space of the process, hence any attempted references to VOS symbols would
result in a segmentation violation.  The \fIzzstrt\fR procedure, called by
the process main during process startup, opens the shared image file and
maps it into the virtual space of the IRAF program.  Once the IRAF main
prompt appears (when running an IRAF process standalone), all initialization
will have completed.
.PP
Each BIN, if linked with the shared library, will have its own shared image
file \f(CWS\fIn\fP.e\fR.  If the shared image is relinked this file will be
moved to \f(CWS\fIn\fP.e.1\fR and the new shared image will take its place;
any old shared image files should eventually be deleted to save disk space,
once any IRAF processes using them have terminated.   Normally when the
shared image is rebuilt it is not necessary to relink applications programs,
since the transfer vector causes the linked application to be unaffected
by relocation of the shared image functions.
.PP
If the shared image is rebuilt and its version number (the \fIn\fR in
\f(CWS\fIn\fP.e\fR) is incremented, the transfer vector is rebuilt the new
shared image cannot be used with previously linked applications.  These
old applications will still continue to run, however, so long as the older
shared image is still available.  It is common practice to have at least
two shared image versions installed in a BIN directory.
.PP
Further information on the Sun/IRAF shared library facility in given in the
IRAF V2.8 system notes file.  In particular, anyone doing extensive IRAF
based software development should review this material, e.g., to learn how
to debug processes that are linked with the shared image.

.NH 2
Layered software support
.PP
An IRAF installation consists of the core IRAF system and any number of
external packages, or "layered software products".  As the name suggests,
layered software products are layered upon the core IRAF system.  Layered
software requires the facilities of the core system to run, and is portable
to any computer which already runs IRAF.  Any number of layered products can
be installed in IRAF to produce the IRAF system seen by the user at a
given site.
.PP
The support provided by IRAF for layered software is essentially the same as
that provided for maintaining the core IRAF system itself (the core system
is a special case of a layered package).  Each layered package (usually this
refers to a suite of subpackages) is a system in itself, similar in
structure to the core IRAF system.  Hence, there is a LIB, one or more BINs,
a help database, and all the sources and runtime files.  A good example of
an external package is the NOAO package.  Except for the fact that NOAO is
rooted in the IRAF directories, NOAO is equivalent to any other layered
product, e.g., STSDAS, TABLES, XRAY, CTIO, NSO, ICE, GRASP, NLOCAL, STEWARD,
and so on.  In general, layered products should be rooted somewhere outside
the IRAF directory tree to simplify updates.

.NH 2
Software management tools
.PP
IRAF software management is performed with a standard set of tools,
consisting of the tasks in the SOFTOOLS package, plus the host system
editors and debuggers.  Some of the most important and often used tools for
IRAF software development and software maintenance are the following.
.sp
.RS
.IP \f(CWmkhelpdb\fP 20
Updates the HELP database of the core IRAF system or an external package.
The core system, and each external package, has its own help database.
The help database is the machine independent file \f(CWhelpdb.mip\fR in the
package library (LIB directory).  The help database file is generated with
\fImkhelpdb\fR by compiling the \f(CWroot.hd\fR file in the same directory.
.IP \f(CWmkpkg\fP 20
The "make-package" utility.  Used to make or update package trees.
Will update the contents of the current directory tree.  When run at
the root iraf directory, updates the full IRAF system; when run at the
root directory of an external package, updates the external package.
Note that updating the core IRAF system does not update any external
packages (including NOAO).  When updating an external package, the
package name must be specified, e.g., "\fImkpkg -p noao\fR".
.IP \f(CWrmbin\fP 20
Descends a directory tree or trees, finding and optionally listing or
deleting all binary files therein.  This is used, for example, to strip
the binaries from a directory tree to leave only sources, to force
\fImkpkg\fR to do a full recompile of a package, or to locate all the
binaries files for some reason.  IRAF has its own notion of what a binary
file is.  By default, files with the "known" file extensions
(.[aoe], .[xfh] etc.) are classified as binary or text
(machine independent) files immediately,
while a heuristic involving examination of the file data
is used to classify other files.  Alternatively, a list of file extensions
to be searched for may optionally be given.
.IP \f(CWrtar,wtar\fP 20
These are the portable IRAF tarfile writer (\fIwtar\fR) and reader
(\fIrtar\fR).  About the only reasons to use these with the UNIX versions of
IRAF are if one wants to move only the machine independent or source files
(\fIwtar\fR, like \fIrmbin\fR, can discriminate between machine generated
and machine independent files), or if one is importing files written to a
tarfile on a VMS/IRAF system, where the files are blank padded and the
trailing blanks need to be stripped with \fIrtar\fR.
.IP \f(CWxc\fP 20
The X (SPP) compiler.  This is analogous to the UNIX \fIcc\fR except
that it can compile ".x" or SPP source files, knows how to link with the
IRAF system libraries and the shared library, knows how to read the
environment of external packages, and so on.
.RE
.sp
.PP
The SOFTOOLS package contains other tasks of interest, e.g., a program
\fImktags\fR for making a tags file for the \fIvi\fR editor, a help
database examine tool, and other tasks.  Further information on these
tasks is available in the online help pages.

.NH 2
Modifying and updating a package
.PP
IRAF applications development is most conveniently performed from within the
IRAF environment, since testing must be done from within the environment.
The usual  edit-compile-test development cycle is illustrated below.  This
takes place within the \fIpackage directory\fR containing all the files
specific to a given package.
.RS
.IP \(bu
Edit one or more source files.
.IP \(bu
Use \fImkpkg\fR to compile any modified files, or files which include a
modified file, and relink the package executable.
.IP \(bu
Test the new executable.
.RE
.PP
The mkpkg file for a package can be written to do anything,
but by convention the following commands are usually provided.
.sp
.RS 
.IP "\f(CWmkpkg\fP" 20
The \fImkpkg\fR command with no arguments does the default mkpkg operation;
for a subpackage this is usually the same as \fImkpkg relink\fR below.  For
the root mkpkg in a layered package it udpates the entire layered package.
.IP "\f(CWmkpkg libpkg.a\fP" 20
Updates the package library, compiling any files which have been modified or
which reference include files which have been modified.  Private package
libraries are intentionally given the generic name libpkg.a to symbolize
that they are private to the package.
.IP "\f(CWmkpkg relink\fP" 20
Rebuilds the package executable, i.e., updates the package library and
relinks the package executable.  By convention, this is the file
xx_\fIpkgname\fR.e\fR in the package directory, where \fIpkgname\fR is the
package name.
.IP "\f(CWmkpkg install\fP" 20
Installs the package executable, i.e., renames the xx_foo.e file to x_foo.e
in the global BIN directory for the layered package to which the subpackage
\fIfoo\fR belongs.
.IP "\f(CWmkpkg update\fP" 20
Does everything, i.e., a \fIrelink\fR followed by an \fIinstall\fR.
.RE
.sp
.PP
If one wishes to test the new program before installing it one should do a
\fIrelink\fR (i.e., merely type "mkpkg" since that defaults to relink), then
run the host system debugger on the resultant executable.  The process is
debugged standalone, running the task by giving its name to the standalone
process interpreter.  The CL task \fIdparam\fR is useful for dumping a
task's parameters to a text file to avoid having to answer parameter queries
during process execution.  The LOGIPC debugging facility introduced in V2.10
is also useful for debugging subprocesses.  If the new program is to be
tested under the CL before installation, a \fItask\fR statement can be
interactively typed into the CL to cause the CL to run the "xx_" version of
the package executable, rather than old installed version.
.PP
When updating a package other than in the core IRAF system, the \f(CW-p\fR
flag, or the equivalent \f(CWPKGENV\fR environment variable, must be used to
indicate the system or layered product being updated.  For example, "mkpkg
-p noao update" would be used to update one of the subpackages of the NOAO
layered package.  If the package being updated references any libraries or
include files in \fIother\fR layered packages, those packages must be
indicated with a "-p pkgname" flag as well, to cause the external package to
be searched.
.PP
The CL process cache can complicate debugging and testing if one forgets
that it is there.  When a task is run under the CL, the executing process
remains idle in the CL process cache following task termination.  If a new
executable is installed while the old one is still in the process cache, the
CL will automatically run the new executable (the CL checks the modify date
on the executable file every time a task is run).  If however an executable is
currently running, either in the process cache or because some other user is
using the program, it may not be possible to set debugger breakpoints.
.PP
The IRAF shared image can also complicate debugging, although for most
applications-level debugging the shared library is transparent.  By default
the shared image symbols are included in the symbol table of an output
executable following a link, so in a debug session the shared image will
appear to be part of the applications program.  When debugging a program
linked with the shared library, the process must be run with the \f(CW-w\fR
flag to cause the shared image to be mapped with write permission, allowing
breakpoints to be set in the shared image (that is, you type something like
":r -w" when running the process under the debugger).  Linking with the
\f(CW-z\fR flag will prevent use of the shared image entirely.
.PP
A full description of these techniques is beyond the scope of this manual,
but one need not be an expert at IRAF software development techniques to
perform simple updates.  Most simple revisions, e.g., bug fixes or updates,
can be made by merely editing or replacing the affected files and typing
.XS
cl> mkpkg
.XE
or
.XS
cl> mkpkg update
.XE
to update the package.

.NH 2
Installing and maintaining layered software
.PP
The procedures for installing layered software products are similar to those
used to install the core IRAF system, or update a package.
Layered software may be distributed in source only form, or with binaries;
it may be configured for a single architecture, or may be preconfigured
to support multiple architectures.  The exact procedures to be followed
to install a layered product will in general be product dependent, and should
be documented in the installation guide for the product.
.LP
In brief, the procedure to be followed should resemble the following:
.RS
.IP \(bu
Create the root directory for the new software, somewhere outside the
IRAF directories.
.IP \(bu
Restore the files to disk from a tape or network archive distribution file.
.IP \(bu
Edit the core system file hlib$extern.pkg to "install" the new package in
IRAF.  This file is the sole link between the IRAF core system and the
external package.
.IP \(bu
Configure the package BIN directory or directories, either by restoring
the BIN to disk from an archive file, or by recompiling and relinking the
package with \fImkpkg\fR.
.RE
.LP
As always, there are some little things to watch out for.
When using \fImkpkg\fR on a layered product, you must give the name
of the system being operated upon, e.g.,
.XS
cl> mkpkg -p foo update
.XE
where \fIfoo\fR is the system or package name, e.g., "noao", "local", etc.
The \f(CW-p\fR flag can be omitted by defining \f(CWPKGENV\fR in your
UNIX environment, but this only works for updates to a single package.
.PP
An external system of packages may be configured for multiple architecture
support by repeating what was done for the core system.  One sets up several
BIN directories, one for each architecture, named \f(CWbin.\fIarch\fR, where
\fIarch\fR is "sparc", "ddec", "rs6000", etc.  These directories, or
symbolic links to the actual directories, go into the root directory of the
external system.  A symbolic link \f(CWbin\fR pointing to an empty directory
bin.generic, and the directory itself, are added to the system's root
directory.  The system is then stripped of its binaries with \fIrmbin\fR, if
it is not already a source only system.  Examine the file zzsetenv.def in
the layered package LIB directory to verify that the definition for the
system BIN (which may be called anything) includes the string "(arch)",
e.g.,
.XS
set noaobin = "noao$bin(arch)/"
.XE
.LP
The binaries for each architecture may then be generated by configuring the
system for the desired architecture and running \fImkpkg\fR to update the
binaries, for example,
.XS
cl> cd foo
cl> mkpkg sparc
cl> mkpkg -p foo update >& spool &
.XE
where \fIfoo\fR is the name of the system being updated.  If any questions
arise, examination of a working example of a system configured for multiple
architecture support (e.g., the NOAO packages) may reveal the answers.
.PP
Once installed and configured, a layered product may be deinstalled merely
by archiving the package directory tree, deleting the files, and commenting
out the affected lines of hlib$extern.pkg.  With the BINs already configured
reinstallation is a simple matter of restoring the files to disk and editing
the extern.pkg file.

.NH 2
Configuring a custom LOCAL package
.PP
Anyone who uses IRAF enough will eventually want to add their own software
to the system, by copying and modifying the distributed versions of
programs, by obtaining and installing isolated programs written elsewhere,
or by writing new programs of their own.  A single user can do this by
developing software for their own personal use, defining the necessary
\fItask\fR statements etc.  to run the software in their personal login.cl
or loginuser.cl file.  To go one step further and install the new software
in IRAF so that it can be used by everyone at a site, one must configure a
custom local package.
.PP
The procedures for configuring and maintaining a custom LOCAL package are
similar to those outlined in \(sc3.5 for installing and maintaining
layered software, since a custom LOCAL will in fact be a layered software
product, possibly even something one might want to export to another site
(although custom LOCALs may contain non-portable or site specific software).
.PP
To make a custom local you make a copy of the "template local" package
(iraf$local) somewhere outside the IRAF directory tree, change the name
to whatever you wish to call the new layered package, and install it as
outlined in \(sc3.5.  The purpose of the template local is to provide the
framework necessary for a external package; a couple of simple tasks are
provided in the template local to serve as examples.  Once you have
configured a local copy of the template local and gotten it to compile and
link, it should be a simple matter to add new tasks to the existing
framework.

.NH 2
Updating the full IRAF system
.PP
This section will describe how to recompile or relink IRAF.  Before we get
into this however, it should be emphasized that \fImost users will never
need to recompile or relink IRAF\fR.  In fact, this is not something that
one should attempt lightly - don't do it unless you have some special
circumstance which requires a custom build of the system (such as a port).
Even then you might want to set up a second copy of IRAF to be used for the
experiment, keeping the production system around as the standard system.  If
you change the system it is a good idea to make sure that you can undo the
change.
.PP
While the procedure for building IRAF is straightforward, it is easy to make
a mistake and without considerable knowledge of IRAF it may be difficult to
recover from such a a mistake (for example, running out of disk space during
a build, or an architecture mismatch resulting in a corrupted library or
shared image build failure).  More seriously, the software - the host
operating system, the host Fortran compiler, the local system configuration,
and IRAF - is changing constantly.  A build of IRAF brings all these things
together at one time, and every build needs to be independently and
carefully tested.  An OS upgrade or a new version of the Fortran compiler
may not yet be supported by the version of IRAF you have locally.  Any
problems with the host system configuration can cause a build to fail, or
introduce bugs.  For example, systems which support multiple Fortran
compilers or which require the user to install and configure the compiler
are a common source of problems.
.PP
The precompiled binaries we ship with IRAF have been carefully prepared and
tested, usually over a period of months prior to a major release.  They are
the same as are used at NOAO and at most IRAF sites, so even if there are
bugs they will likely have already been seen elsewhere and a workaround
determined.  If the bugs are new then since we have the exact same IRAF
system we are more likely to be able to reproduce and fix the bug.  Often
the bug is not in the IRAF software at all but in the host system or IRAF
configuration.  As soon as an executable is rebuilt (even something as
simple as a relink) you have new, untested, software.
.NH 3
The BOOTSTRAP
.PP
To fully build IRAF from the sources is a three step process.  First the
system is "bootstrapped", which builds the host system interface (HSI)
executables.  A "sysgen" of the core system is then performed; this compiles
all the system libraries and builds the core system applications.  Finally,
the bootstrap is then repeated, to make use of some of the functions from
the IRAF libraries compiled in step two.
.PP
To bootstrap IRAF, login as IRAF and enter the commands shown below.
This takes a while and generates a lot of output, so the output should be
spooled in a file.  Here, \fIarch\fR refers to the IRAF architecture you
wish to build for.
.XS
% cd $iraf
% mkpkg \fIarch\fP
% cd $iraf/unix
% reboot >& spool &
.XE
.PP
There are two types of bootstrap, the initial bootstrap starting from a
source only system, called the NOVOS bootstrap, and the final or VOS
bootstrap, performed once the IRAF system libraries \f(CWlibsys.a\fR and
\f(CWlibvops.a\fR exist.  The bootstrap script \fIreboot\fR will
automatically determine whether or not the VOS libraries are available and
will perform a NOVOS bootstrap if the libraries cannot be found.  It is
important to restore the desired architecture before attempting a
bootstrap, as otherwise a NOVOS bootstrap will be performed.
.NH 3
The SYSGEN
.PP
By sysgen we refer to an update of the core IRAF system - all of the files
comprising the runtime system, excluding the HSI which is generated by the
bootstrap.  On a source only system, the sysgen will fully recompile the
core system, build all libraries and applications, and link and install the
shared image and executables.  On an already built system, the sysgen
scans the full IRAF directory tree to see if anything is out of date,
recompiles any files that need it, then relinks and installs new executables.
.PP
To do a full sysgen of IRAF one merely runs \fImkpkg\fR at the IRAF root.
If the system is configured for multiple architecture support one must
repeat the sysgen for each architecture.  Each sysgen builds or updates a
single BIN directory.  Since a full sysgen takes a long time and generates a
lot of output which later has to be reviewed, it is best to run the job in
batch mode with the output redirected.  For example to update the ddec
binaries on a Decstation:
.XS
% cd $iraf
% mkpkg ddec
% mkpkg >& spool &
.XE
To watch what is going on after this command has been submitted and while
it is running, try
.XS
% tail -f spool
.XE
Sysgens are restartable, so if the sysgen aborts for any reason, simply fix
the problem and start it up again.  Modules that have already been compiled
should not need to be recompiled.  How long the sysgen takes depends upon
how much work it has to do.  The worst case is if the system and
applications libraries have to be fully recompiled.  If the system libraries
already exist they will merely be updated.  Once the system libraries are up
to date the sysgen will rebuild the shared library if any of the system
libraries involved were modified, then the core system executables will be
relinked.
.PP
A full sysgen generates a lot of output, too much to be safely reviewed for
errors by simply paging the spool file.  Enter the following command to
review the output (this assumes that the output has been saved in a file
named "spool").
.XS
% mkpkg summary
.XE
It is normal for a number of compiler messages warning about assigning
character data to an integer variable to appear in the spooled output
if the full system has been compiled.  There should be no serious error
messages if a supported and tested system is being recompiled.
.PP
The above procedure only updates the core IRAF system.  To update a layered
product one must repeat the sysgen process for the layered system.  For
example, to update the sparc binaries for the NOAO package:
.XS
% cd $iraf/noao
% mkpkg ddec
% mkpkg -p noao >& spool &
.XE
This must be repeated for each supported architecture.  Layered systems are
independent of one another and hence must be updated separately.
.PP
To force a full recompile of the core system or a layered package, one can
use \fIrmbin\fR to delete the objects, libraries, etc. scattered throughout
the system, or do a "mkpkg generic" and then delete the \f(CWOBJS.arc.Z\fR
file in the BIN one wishes to regenerate (the latter approach is probably
safest).
.PP
A full IRAF core system sysgen currently takes anywhere from 3 to 30 hours,
depending upon the system (e.g. from 30 hours on a VAX 11/750, to 3 hours on
a big modern server).  On most systems a full sysgen is a good job to run
overnight.
.PP
.NH 3
Localized software changes
.PP
The bootstrap and the sysgen are unusual in that they update the entire
HSI, core IRAF system, or layered package.  Many software changes are more
localized.  If only a few files are changed a sysgen will pick up the changes
and update whatever needs to be updated, but for localized changes a sysgen
really does more than it needs to (if the changes are scattered all over
the system an incremental sysgen-relink will still be best).
.PP
To make a localized change to a core system VOS library and update the
linked applications to reflect the change all one really needs to do is
change the desired source files, run \fImkpkg\fR in the library source
directory to compile the modules and update the affected libraries, and then
build a new IRAF shared image (this assumes that the changes affect only the
libraries used to make the shared image, i.e., libsys, libex, libvops, and
libos).  Updating only the shared image, without relinking all the
applications, has the advantage that you can put the runtime system back the
way it was by just swapping the old shared image back in - a single file.
.PP
For example, assume we want to make a minor change to some files in the VOS
interface IMIO, compiling for the sparc architecture on SunOS, which uses a
shared library.  We could do this as follows (this assumes that one is
logged in as IRAF and that the usual IRAF environment is defined).
.XS
% whoami
iraf
% cd $iraf
% mkpkg sparc
% cd imio
     \fR(edit the files)\fP
% mkpkg                          \fR# update IMIO libraries (libex)\fP
%
% cd $iraf/bin.sparc             \fR# save copy of old shared image\fP
% cp S6.e S6.e.V210
%
% cd shlib
% tar -cf ~/shlib.tar .          \fR# backup shlib just in case\fP
% mkpkg update                   \fR# make and install new shared image\fP
.XE
.PP
If IRAF is not configured with shared libraries, one must relink the full
IRAF system and all layered packages for the change to take effect.  This 
is done by running \fImkpkg\fR at the root of the core system and each layered
package.  For example, on an IBM RS/6000,
.XS
% whoami
iraf
% cd $iraf
% mkpkg rs6000
% cd imio
     \fR(edit the files)\fP
% cd iraf
% mkpkg                         \fR# update the core system\fP
%
% cd noao
% mkpkg rs6000
% mkpkg -p noao                 \fR# update the NOAO packages\fP
.XE
.LP
and so on, for each layered package.
.PP
Changing applications is even easier.  Ensure that the system architecture
is set correctly (i.e. "mkpkg \fIarch\fR" at the iraf or layered package root),
edit the affected files in the package source directory, and type "mkpkg -p
<pkgname> update" in the root directory of the package being edited.  This
will compile any modified files, and link and install a new executable.
You can do this from within the CL and immediately run the revised program.
.PP
We should emphasize again that, although we document the procedures for
making changes to the software here, to avoid introducing bugs we do not
recommend changing any of the IRAF software except in unusual (or at least
carefully controlled) circumstances.  To make custom changes to an
application, it is best to make a local copy of the full package somewhere
outside the standard IRAF system.  If changes are made to the IRAF system
software it is best to set up an entire new copy of IRAF on a machine
separate from the normal production installation, so that one can experiment
at will without affecting the standard system.  An alternative which does
not require duplicating the full system is to use the \f(CWIRAFULIB\fR
environment variable.  This can be used to safely experiment with custom
changes to the IRAF system software outside the main system; IRAFULIB lets
you define a private directory to be searched for IRAF global include files,
libraries, executables, etc., allowing you to have your own private versions
of any of these.  See the system notes files for further information on how
to use IRAFULIB.

.NH
Graphics and Image Display
.PP
IRAF itself is device and window system independent, hence it can be used
with any windowing system such as X11 or SunView, or with hardware graphics
and display devices.  Nowadays most people will be running IRAF on a UNIX
workstation under X11.  At the time that this is being written, IRAF is most
commonly run under X11 using \fIxterm\fR for graphics and \fIsaoimage\fR for
image display.  Binaries for these applications are included in the IRAF
distribution if not already provided with the window system software on the
host system.  New graphics and image display clients are being developed for
use with IRAF running under X11; contact the IRAF group for further
information on the availability of these products.
.NH 2
Using the workstation with a remote compute server
.PP
A common mode of operation with a workstation is to run IRAF under a window
system directly on the workstation which runs IRAF, accessing files either
on a local disk, or on a remote disk via a network interface (NFS, IRAFKS,
etc.).  It is also possible, however, to run the window system on the
workstation, but run IRAF on a remote node, e.g., some powerful compute
server such as a large UNIX server, a large VAX, vector minisupercomputer,
supercomputer, etc., possibly quite some distance away.  This is done by
logging onto the workstation, starting up the window system, logging onto
the remote machine with \fIrlogin\fR, \fItelnet\fR, or whatever, and
starting up IRAF on the remote node.
.PP
If X11 is running on the local workstation as well as on the remote system,
and one's favorite X11 client it installed on the remote system, then the
networking support built into X11 can be used to display and plot remotely.
This is not always possible however.  If the necessary X11 clients are not
available on the remote system or the networking connection does not support
X11, it is still possible to work remotely using the networking capabilities
built into IRAF, provided one is already running IRAF on the remote node.
.LP
After IRAF comes up one need only type
.XS
cl> stty xterm
cl> reset node = \fIhostname\fP
.XE
to tell the remote IRAF that it is talking to an xterm window (for example)
and that the image display is on the network node \fIhostname\fR.

.NH
Interfacing New Graphics Devices
.PP
There are three types of graphics devices that concern us here.
These are the graphics terminals, graphics plotters, and image displays.
.NH 2
Graphics terminals
.PP
The IRAF system as distributed is capable of talking to just about any
conventional graphics terminal or terminal emulator, using the \fIstdgraph\fR
graphics kernel supplied with the system.  All one need do to interface to a
new graphics terminal is add new graphcap and termcap entries for the device.
This can take anywhere from a few hours to a few days, depending on one's
level of expertise, and the characteristics of the device.  Be sure to check
the contents of the dev$graphcap file to see if the terminal is already
supported, before trying to write a new entry.  Useful documentation for
writing graphcap entries is the GIO reference manual and the HELP pages for
the \fIshowcap\fR and \fIstty\fR tasks (see \(sc2.2.6).  Assistance with
interfacing new graphics terminals is available via the IRAF Hotline.
.NH 2
Graphics plotters
.PP
The current IRAF system comes with several graphics kernels used to drive
graphics plotters.  The standard plotter interface the SGI graphics kernel,
which is interfaced as the tasks \fIsgikern\fR and \fIstdplot\fR in the
PLOT package.  Further information on the SGI plotter interface is given in
the paper \fIThe IRAF Simple Graphics Interface\fR, a copy of which is
included with the IRAF installation kit.
.PP
SGI device interfaces for most plotter devices already exist, and adding
support for new devices is straightforward.  Sources for the SGI device
translators supplied with the distributed system are maintained in the
directory iraf/unix/gdev/sgidev.  NOAO serves as a clearinghouse for new SGI
plotter device interfaces; contact us if you do not find support for a local
plotter device in the distributed system, and if you plan to implement a new
device interface let us know so that we may help other sites with the same
device.
.PP
The older NCAR kernel is used to generate NCAR metacode and can be
interfaced to an NCAR metacode translator at the host system level to get
plots on devices supported by host-level NCAR metacode translators.  The
host level NCAR metacode translators are not included in the standard IRAF
distribution, but public domain versions of the NCAR implementation for UNIX
systems are widely available.  A site which already has the NCAR software
may wish to go this route, but the SGI interface will provide a more
efficient and simpler solution in most cases.
.PP
The remaining possibility with the current system is the \fIcalcomp\fR kernel.
Many sites will have a Calcomp or Versaplot library (or Calcomp compatible
library) already available locally.  To make use of such a library to get
plotter output on any devices supported by the interface, one may copy
the library to the hlib directory and relink the Calcomp graphics
kernel.
.PP
A graphcap entry for each new device will also be required.  Information on
preparing graphcap entries for graphics devices is given in the GIO design
document, and many actual working examples will be found in the graphcap
file.  The best approach is usually to copy one of these and modify it.
.NH 2
Image display devices
.PP
The standard image display facility for a Sun workstation running the
SunView window system is \fIimtool\fR.  Image display under the
MIT X window system is also available using the \fIsaoimage\fR display
server.  This was developed for IRAF by SAO; distribution kits are available
from the IRAF network archive.  At the time that this was written new
X11 based image display clients were being developed for or interfaced to
IRAF by several sites.  Eventually, there will be a range of image display
clients to choose from and people will use the tool best suited to the
type of data analysis they are doing.
.PP
Some interfaces for hardware image display devices are also available,
although a general display interface is not yet included in the system.
Only the IIS model 70 and 75 are current supported by NOAO.  Interfaces for
other devices are possible using the current datastream interface, which is
based on the IIS model 70 datastream protocol with extensions for passing
the WCS, image cursor readback, etc. (see the ZFIOGD driver in unix/gdev).
This is how all the current displays, e.g., imtool and ximage, and the IIS
devices, are interfaced, and there is no reason why other devices could not
be interfaced to IRAF via the same interface.  Eventually this prototype
interface will be obsoleted and replaced by a more general interface.

.NH
Host System Requirements
.PP
Any modern host system capable of running UNIX should be capable of running
IRAF as well.  IRAF is supported on all the more popular UNIX platforms,
as well as on other operating systems such as VMS.
.PP
A typical small system is a single workstation with a local disk.  In a
typical large installation there will be one or more large central compute
servers, each with several Gb of disk and many Mb of RAM, networked to a
number of personal or public workstations.  For scientific use, a megapixel
color screen is desirable.
.NH 2
Memory requirements
.PP
The windowing systems used in these workstations tend to be very memory
intensive; the typical screen with ten or so windows uses a lot of memory.
Interactive performance will suffer greatly if the system pages a lot.
Fortunately, memory is becoming relatively cheap.  Typical workstation
memory sizes in 1992 range from 16 to 32 Mb.  Servers will have several
times that.
.NH 2
Disk requirements
.PP
IRAF itself requires anywhere from 60 to 150 Mb of memory depending upon
whether the system is stripped, on the size of the binaries, and on how many
architectures are supported.  Since IRAF is an image processing system,
usually the disk requirements of the data will vastly outstrip those of IRAF
itself.  The amount of space needed for the data to be processed varies
greatly and will depend upon the type of data being processed.  A useful
system requires from several hundred Mb to 1 Gb of disk space.

.SH
Appendix A.  The IRAF Directory Structure
.PP
The main branches of the IRAF directory tree are summarized below.
Beneath the directories shown are some 400 subdirectories, the largest
directory trees being \f(CWsys\fR, \f(CWpkg\fR, and \f(CWnoao\fR.
The entire contents of all directories other than \f(CWunix\fR, \f(CWlocal\fR,
and \f(CWdev\fR are fully portable, and are identical in all installations
of IRAF sharing the same version number.
.XS
bin        \fR- the IRAF BIN directories\fP
dev        \fR- device tables (termcap, graphcap, etc.)\fP
doc        \fR- assorted IRAF manuals\fP
lib        \fR- the system library; global files\fP
local      \fR- iraf login directory; locally added software\fP
math       \fR- sources for the mathematical libraries\fP
noao       \fR- packages for NOAO data reduction\fP
pkg        \fR- the IRAF applications packages\fP
sys        \fR- the virtual operating system (VOS)\fP
unix       \fR- the UNIX host system interface (HSI = kernel + bootstrap utilities)\fP
.XE
.LP
The contents of the \f(CWunix\fR directory (host system interface) are
as follows:
.XS
as         \fR- assembler sources\fP
bin        \fR- the HSI BIN directories\fP
boot       \fR- bootstrap utilities (mkpkg, rtar, wtar, etc.)\fP
gdev       \fR- graphics device interfaces (SGI device translators)\fP
hlib       \fR- host dependent library; global files\fP
os         \fR- OS interface routines (UNIX/IRAF kernel)\fP
reboot     \fR- executable script run to reboot the HSI\fP
shlib      \fR- shared library facility sources\fP
sun        \fR- gterm and imtool sources (SunView)\fP
x11        \fR- saoimage and other X11 sources\fP
.XE
.PP
If you will be working with the system much at the system level, it will be
well worthwhile to spend some time exploring these directories and gaining
familiarity with the system.