++++++++++++++ INSTALLING ZSH ++++++++++++++ This file is divided into two parts: making and installing the shell, a note on the script run to set up the environment for new users, and a description of various additional configuration options. You should have a look at the items in the second and third parts before following the instructions in the first. ===================== MAKING AND INSTALLING ===================== Dependencies ------------ The main shell can be compiled with tools available on a typical Unix-like system including development packages. However, the documentation requires various additional tools that are not so standard: - YODL, for turning the documentation source files (.yo) into manual pages and TeXinfo source for producing PDF and .info formats; - Perl and interactive manual tools, for turning the zshbuiltins(1) manual into help files that can be displayed by the run-help function. For example, the colcrt or col tool, not typically required for installing software, must be available. To avoid needing these, the source distribution is supplemented by a documentation distribution that may be downloaded from the same place as the source. This provides fully generated documentation with references to files in the default locations. Check MACHINES File ------------------- Check the file MACHINES in the top directory to see the architectures that zsh is known to compile on, as well as any special instructions for your particular architecture. Most architectures will not require any special instructions. Pre-configuration ----------------- If you are using a normal source release, skip this section. If the `configure' script does not already exist -- e.g., if you've got a snapshot of the bare sources just checked out from a git repository -- some things need to be built before the configuration can proceed. Run the script `./Util/preconfig' to do this. Configuring Zsh --------------- To configure zsh, from the top level directory, do the command: ./configure Configure accepts several options (explained below). To display currently available options, do the command: ./configure --help Many of the interesting configuration options can be added after running configure by editing the user configuration section of config.h and the top level Makefile. However, see the end of this file for a list of features configurable on the command line. Dynamic loading --------------- Zsh has support for dynamically loadable modules. This is now enabled by default; to disable it, run configure with the --disable-dynamic option. Note that dynamic loading does not work on all systems. On these systems this option will have no effect. When dynamic loading is enabled, major parts of zsh (including the Zsh Line Editor) are compiled into modules and not included into the main zsh binary. Zsh autoloads these modules when they are required. This means that you have to execute make install.modules before you try the newly compiled zsh executable, and hence also the install paths must be correct. The installation path for modules is EPREFIX/lib/zsh/, where EPREFIX defaults to PREFIX unless given explicitly, and PREFIX defaults to /usr/local. See the end of this file for options to configure to change these. Adding and removing modules --------------------------- The zsh distribution contains several modules, in the Src/Builtins, Src/Modules and Src/Zle directories. If you have any additional zsh modules that you wish to compile for this version of zsh, create another subdirectory of the Src directory and put them there. You can create as many extra subdirectories as you need, but currently configure will only search in immediate subdirectories of Src. The subdirectories must be actual directories; symbolic links will not work. You will then need to rerun configure; the easiest way is to run `config.status --recheck' from the top-level build directory which retains the existing configuration as much as possible. The key to the module system is the file config.modules, created in the configuration process. In the normal case that dynamic loading is available, all modules relevant to your configuration will be compiled and installed as separate files, so unless you want the modules to be loaded by default you don't need to do anything. For a non-dynamic zsh, the default is to compile the complete, compctl, zle, computil, complist, sched, parameter, zleparameter and rlimits modules into the shell, and you will need to edit config.modules to make any other modules available. If you wish to change the configuration, here is how config.modules works. Each module has a line in the file. Be careful to retain the (strict) format for lines in the file: link - `dynamic', if the module is to be dynamically linked -- meaningless if this is not available on your system. `static' if the module is to be linked directly into the executable. `no' if the module is not to be linked at all. In this case it will not even be compiled. load - `yes' if the module is to be visible to the user. This will make builtins, parameters etc. visible to the user without any need to use the zmodload builtin. `no' if an explicit zmodload command is to be required to load the utilities in the module. Note that this applies both to statically and dynamically linked modules. auto - `yes' if the entry is to be regenerated whenever configure is run. `no' if you wish to retain your hand-edited version. Do not edit the entry for the pseudo-module zsh/main (apart from the `functions=' part) as this is the main shell. After you have edited this file, run `make prep' in the Src subdirectory. Note that the modules depending on zle or complete (e.g.: complist and deltochar) cannot be loaded dynamically on systems which do not allow symbols in one dynamically loaded library to be visible from another; this is true, for example, of version 4 of SunOS. The most convenient workaround is to compile zle and complete into the base executable by setting their `link' entries in config.modules to `static' as described above. Compiler Options or Using a Different Compiler ---------------------------------------------- By default, configure will use the "gcc" compiler if found. You can use a different compiler, or add unusual options for compiling or linking that the "configure" script does not know about, by either editing the user configuration section of the top level Makefile (after running configure) or giving "configure" initial values for these variables by setting them in the environment. Using a Bourne-compatible shell (such as sh,ksh,zsh), you can do that on the command line like this: CC=c89 ./configure --enable-cflags=-O2 --enable-libs=-lposix This is almost equivalent to CC=c89 CFLAGS=-O2 LIBS=-lposix ./configure but has the advantage that the CFLAGS and LIBS variables are remembered if the configuration is recreated by means of `config.status --recheck' (this happens automatically if certain configuration files change). You can set the make variables CFLAGS, CPPFLAGS, LDFLAGS and LIBS in this way, however CC must appear as shown. If you are configuring from a csh-derived shell, you may need to use the "env" program: env CC=c89 ./configure --enable-cflags=-O2 --enable-libs=-lposix. You can override the variables directly when running `make': make CFLAGS=-g However, these will not be passed down via `config.status --recheck'. Check Generated Files --------------------- Configure will probe your system and create a "config.h" header file. You should check the user configuration section at the beginning of this include file. You should also examine the values (determined by configure) of HOSTTYPE, OSTYPE, MACHTYPE, and VENDOR to make sure they are correct. The value of these #defines's is used only to initialize the corresponding default shell parameters. Since these shell parameters are only for informational purposes, you can change them to whatever you feel is appropriate. Also, configure will create a Makefile in the top level directory as well as in the various subdirectories. You should check the user configuration section of the top level Makefile. Compiling Zsh ------------- After configuring, to build zsh, execute the command: make It's then a good idea to check that your build is working properly: make check If you have trouble with a particular test, you can run it separately: make TESTNUM=C02 check The TESTNUM value can be a single test number, as above, or a letter to run an entire category of tests: make TESTNUM=Y check See Test/README for a list of test categories. Installing Zsh -------------- If no make/compilation errors occur, then execute the command make install to install all the necessary files except for the info files. Alternatively, you can install the various parts in separate stages. To install the zsh binary, execute the command: make install.bin Any previous copy of zsh will be renamed "zsh.old" To install the dynamically-loadable modules, execute the command: make install.modules Note that this is required for the shell to operate properly if dynamic loading is enabled. To install the zsh man page, execute the command: make install.man To install all the shell functions which come with the distribution, execute the command: make install.fns To install the zsh info files (this must be done separately), execute the command: make install.info If the programme install-info is available, "make install.info" will insert an entry in the file "dir" in the same directory as the info files. Otherwise you will have to edit the topmost node of the info tree "dir" manually in order to have the zsh info files available to your info reader. Building Zsh On Additional Architectures ---------------------------------------- To build zsh on additional architectures, you can do a "make distclean". This should restore the zsh source distribution back to its original state. You can then configure zsh as above on other architectures in which you wish to build zsh. Or alternatively, you can use a different build directory for each architecture. Using A Different Build Directory --------------------------------- You can compile the zsh in a different directory from the one containing the source code. Doing so allows you to compile it on more than one architecture at the same time. To do this, you must use a version of "make" that supports the "VPATH" variable, such as GNU "make". "cd" to the directory where you want the object files and executables to go and run the "configure" script. "configure" automatically checks for the source code in the directory that "configure" is in. For example, cd /usr/local/SunOS/zsh /usr/local/src/zsh-3.0/configure make Note that this is mutually exclusive with using the source directories as make can become confused by build files created in the source directories. Writing third-party autoloadable functions ------------------------------------------ Third-party autoloadable functions, including but not limited to completion functions, should be installed into the share/zsh/site-functions/ directory under the respective installation prefix. That would typically be written as $(DESTDIR)$(PREFIX)/share/zsh/site-functions/ in a makefile. If the third-party tool's $(PREFIX) is not the same as zsh's prefix, then that directory should be added to $fpath in zsh's initialization files. ================================ AUTOMATIC NEW USER CONFIGURATION ================================ In the default configuration, the shell comes with a system based around the zsh/newuser add-on module that detects when a user first starts the shell interactively and has no initialisation files (.zshenv, .zshrc, .zprofile or .zlogin). The shell then executes code in the file scripts/newuser in the shared library area (by default /usr/local/share/zsh//scripts/newuser). This feature can be turned off simply by removing or renaming this script. This is the recommended way of disabling the features as the function zsh-newuser-install (see below) remains available for users who wish to run it. The module can be removed entirely from the configured shell by editing the line starting "name=zsh/newuser" in the config.modules file, which is generated in the top level distribution directory during configuration: change the line to include "link=no auto=no". The supplied script executes the function supplied as Functions/Newuser/zsh-newuser-install, which is installed to the user's function path when the module is installed. The function can be run by hand after being marked for autoload. This is documented in the zshcontrib manual or in the info node `User Configuration Functions'. zsh-newuser-install is currently under development. It is probably preferable for administrators who wish to customize the system their own way to edit the newuser script in scripts/newuser; for example, this script could copy skeleton files into place safe in the knowledge that the files don't yet exist. Also, as there is currently no internationalization support, administrators of sites with users who mostly do not speak English may wish not to install the zsh/newuser module. ===================== CONFIGURATION OPTIONS ===================== Modified versions of zsh ------------------------ If you are making local modifications to zsh, you are strongly advised to configure with the option --enable-custom-patchlevel="" so that the variable $ZSH_PATCHLEVEL indicates this is not a standard version of the shell. The argument is arbitrary, but should indicate the entity that is customizing the shell, for example the OS vendor or distributor. Multibyte Character Support --------------------------- Support for multibyte character sets that extend ASCII, such as UTF-8, is now reasonably close to complete, except that combining characters are not handled properly (some assistance with this problem would be appreciated). The configuration script should turn on multibyte support on all systems where it can be compiled successfully. The support can be explicitly enabled or disabled with --enable-multibyte or --disable-multibyte. The developers are not aware of any need to use --disable-multibyte and this should be reported as a bug. Currently multibyte mode is believed to work on at least the following: - All(?) current GNU/Linux distributions - OS X 10.4.3 (problems have been reported with multibyte characters in HFS file names) - NetBSD 2.0.2 - Solaris 8+ (inputting multibyte characters from the keyboard doesn't work in some installations). - Cygwin (though use of multibyte characters is somewhat non-standard). The corresponding shell option MULTIBYTE is now on by default in all emulation modes when multibyte support is enabled. Turning it off is not recommended unless there is a particular need to examine single bytes regardless of the locale. As the line editor bases its behaviour on the locale regardless of the option (in order to correspond to the displayed character set), the option should be left on during the execution of user-defined editor and completion widgets so that the behaviour corresponds to that of builtin widgets. See chapter 5 in the FAQ for some notes on multibyte input. Terminal Handling ----------------- Historically, several different libraries have provided the features the shell needs to provide output to the terminal. The most common have been termcap, which is now largely outmoded, and curses, which supersedes termcap and typically contains the same features as well as others. configure will search for an appropriate library; the default search order is "ncursesw tinfo termcap ncurses curses" except on HP-UX ("Hcurses ncursesw ncurses curses termcap") and Solaris ("ncursesw ncurses curses termcap"). Note that even though termcap is searched before traditional forms of curses zsh tries to make features from curses available and if the curses library contains both curses and termcap features, as is normal, the curses variant is used. ncurses is a newer version of curses and tinfo is related to it. The library ncursesw is a variant of ncurses that supports wide characters. zsh attempts to use this to provide functions needed by the zsh/curses module; depending on the configuration, the main shell may not require the additional functions. As the integration of wide character support into ncurses is continuing, it is possible that on some systems attempting to use ncursesw may cause problems during building. If so, please report this to the developers at zsh-workers@zsh.org and attempt to recompile with --with-term-lib="tinfo termcap ncurses curses" (see below). Note that use of ncurses requires the header ncurses.h, so this needs to be in the include path. configure will not search for ncurses or ncursesw unless this is the case. If you have installed ncurses.h in a non-standard place you may need to pass CPPFLAGS=-I/usr/local/include (or wherever the header is found) to configure. Similarly, you may need to pass LDFLAGS=-L/usr/local/lib (or wherever) in order to find the library. You may also need to add headers if you need to include the file term.h from an installation of an ncurses variant instead of from the system directories. For example, CPPFLAGS=-I/usr/local/include/ncursesw will force configure to search for term.h in the given directory. On some systems a suitable development package with a name such as curses-devel or ncurses-devel needs to be installed before zsh can be compiled. This is likely to be contained on any installation media, or available for download. It is highly unlikely that you will need to compile this from scratch. You can tell configure which libraries to search by passing an argument via --with-term-lib. This takes a space-separated list of libraries to try as its argument, so the default is equivalent to --with-term-lib="ncursesw tinfo termcap ncurses curses". It replaces the old option --with-curses-terminfo, which altered the search order but didn't allow an explicit search list to be passed. Memory Routines --------------- Included in this release are alternate malloc and associated functions which reduce memory usage on some systems. To use these, add the option --enable-zsh-mem when invoking "configure". You should check MACHINES to see if there are specific recommendations about using the zsh malloc routines on your particular architecture. Debugging Routines ------------------ You can turn on various debugging options when invoking "configure". To turn on some extra checking in the memory management routines, you can use the following options when invoking "configure". --enable-zsh-mem-warning # turn on warnings of memory allocation errors --enable-zsh-secure-free # turn on memory checking of free() If you are using zsh's memory allocation routines (--enable-zsh-mem), you can turn on debugging of this code. This enables the builtin "mem". --enable-zsh-mem-debug # debug zsh's memory allocators You can turn on some debugging information of zsh's internal hash tables. This enables the builtin "hashinfo". --enable-zsh-hash-debug # turn on debugging of internal hash tables To add some sanity checks and generate debugging information for debuggers you can use the following option. This also disables optimization. --enable-zsh-debug # use it if you want to debug zsh In this mode, zsh may output extra information about internal errors to stderr. The shell variable ZSH_DEBUG_LOG may be set to another file to which errors will be appended. Startup/shutdown files ---------------------- Zsh has several startup/shutdown files which are in /etc by default. This can be overridden using one of the options below when invoking "configure". --enable-etcdir=directory # default directory for global zsh scripts --enable-zshenv=pathname # the full pathname of the global zshenv script --enable-zshrc=pathname # the full pathname of the global zshrc script --enable-zlogin=pathname # the full pathname of the global zlogin script --enable-zprofile=pathname # the full pathname of the global zprofile script --enable-zlogout=pathname # the full pathname of the global zlogout script Any startup/shutdown script can be disabled by giving the --disable-SCRIPTNAME option to "configure". The --disable-etcdir option disables all startup/shutdown files which are not explicitly enabled. The precompiled documentation comes with startup files listed as being in the /etc directory. If yodl is available, running "touch version.yo" and "make" in the Doc directory should produce correctly substituted documentation. (This has to be run by hand to avoid introducing a dependency on yodl on systems where the precompiled documentation is sufficient.) Shell functions --------------- By default, the shell functions which are installed with `make install' or `make install.fns' go into the directory ${datadir}/zsh/functions, which unless you have specified --datadir is the same as ${prefix}/share/zsh/$ZSH_VERSION/functions ($prefix itself defaults to /usr/local, as described below). This directory will also be compiled into the shell as the default directory for the parameters $fpath and $FPATH. You can override it with --enable-fndir=directory; --disable-fndir or --enable-fndir=no will turn off both installation of functions and the setting of a default value for $fpath/$FPATH. Note the presence of $ZSH_VERSION (e.g. `3.1.7') to avoid clashes between versions of zsh. If you only run one version of zsh at once, installing into a common directory such as /usr/local/share/zsh/functions is fine --- note, however, that uninstallation is more likely to create problems in this case. The functions to be installed are controlled by config.modules. These appear at the end of the line after `functions=': note that the rest of the line is taken verbatim as shell command line text, i.e. no quoting is used around the value as a whole and unquoted wildcards will be expanded. To prevent any functions from being installed, either remove the `functions=' entry or delete the rest of the line after it. Functions not specific to a particular module are listed on the zsh/main line. None of these are crucial to shell operation, so you may choose not to install them. For other modules, the functions will be installed if and only if the module itself is installed. This will usually be what you want; in particular, the zsh/complete and zsh/zftp modules are of much less use without the associated functions. The functions listed with zsh/zle are not used by the editor unless you explicitly load them, however. You can also use the configure option --enable-function-subdirs to allow shell functions to be installed into subdirectories of the function directory, i.e. `Base/*' files will be installed into `FNDIR/Base, and so on. This also initialises $fpath/$FPATH appropriately. The option --enable-site-fndir controls whether to create and initialise $fpath to include a directory for site-specific functions. By default this is created in the location ${datadir}/zsh/site-functions, i.e. parallel to the version-specific functions directory, and inserted at the start of the $fpath array on shell startup. This directory will not be affected by `make uninstall' or `make uninstall.fns', although the version-specific directory and its contents will be deleted. The --enable-additional-fpath option may be used to add arbitrary directories to the shell's default $fpath array. This may be useful to have vendor specific function directories available for vendor specific addons. You may add more than one directory this way by listing them with the option separated by commas. The additional directories will be added after the site specific directory (--enable-site-fndir) in the same order in which they are supplied. Function depth -------------- Shell functions may be called recursively. In order to detect infinite recursion the shell has a limit on the depth to which functions may be called: note that this is a single limit for all functions, not a limit for each function called recursively. The default for the limit is 4096. The limit may be altered to the value MAX by passing the option --enable-max-function-depth=MAX to configure. Alternatively, the limit may be disabled with --disable-max-function-depth. However, this is not recommended as it is likely to cause the shell to crash on an infinite recursion. Support for large files and integers ------------------------------------ Some 32-bit systems allow special compilation modes to get around the 2GB file size barrier. This is enabled by default; use --disable-largefile to turn it off. Not all systems recognize the test used by zsh (via the getconf command), so flags may need to be set by hand. On HP-UX 10.20, zsh has been successfully compiled with large file support by configuring with CC="cc -Ae" CPPFLAGS="-D_LARGEFILE_SOURCE -D_FILE64" configure \ --enable-largefile ... Furthermore, use of --enable-largefile will also enable 64-bit arithmetic for shell parameters, and anywhere they are used such as in mathematical formulae. This depends only on the shell finding a suitable 64-bit integer type; it does not require that support for large files is actually enabled. Hence --enable-largefile is useful on many 32-bit systems with a suitable compiler such as gcc. Also note that if `configure' finds out that either of the types off_t or ino_t are 64-bit quantities, but that long integers are only 32 bits, all the above will be enabled automatically. This is necessary to ensure correct handling of these types. None of this is relevant for 64-bit systems; zsh should compile and run without problems if (sizeof(long) == 8). Searching for additional features --------------------------------- Various additional features are turned off by default to avoid compatibility problems. --enable-pcre: Zsh has a module which allows the pcre regular expression library to be used via shell builtins. Compiling this library into the shell with dynamic loading (the default where available) produces a dependency on the library libpcre.so. This is a problem on systems where zsh needs to be available at boot before the directory containing libpcre.so (for example /usr/lib or /usr/local/lib) is mounted. For this reason, pcre support will only be searched for if the option --enable-pcre is passed to configure. (Future versions of the shell may have a better fix for this problem.) --enable-cap: This searches for POSIX capabilities; if found, the `cap' library is available and the shell will use these to determine if the shell is running in some privileged mode. This is turned off by default as on some systems non-standard headers (in particular AIX) are required. A direct fix for that problem would be appreciated. A test for the function tcsetpgrp is turned on by default. The test needs to run the function to determine if the implementation is usable. However, this can cause problems when configure is run without a controlling terminal (eg. from cron). To avoid this, use --with-tcsetpgrp or --without-tcsetpgrp to tell configure whether the function should be used. Options For Configure --------------------- The `configure' program accepts many options, not all of which are useful or relevant to zsh. To get the complete list of configure options, run "./configure --help". The following list should contain most of the options of interest for configuring zsh. Configuration: --cache-file=FILE # cache test results in FILE --help # print a help message --version # print the version of autoconf that create configure --quiet, --silent # do not print `checking...' messages --no-create # do not create output files Directories: --prefix=PREFIX # install host independent files in PREFIX [/usr/local] --exec-prefix=EPREFIX # install host dependent files in EPREFIX [PREFIX] --bindir=DIR # install user executables in DIR [EPREFIX/bin] --infodir=DIR # install info documentation in DIR [PREFIX/info] --mandir=DIR # install man documentation in DIR [PREFIX/man] --srcdir=DIR # find the sources in DIR [configure dir or ..] --datadir=DATADIR # install shared files in DATADIR [PREFIX/share] Features: --enable-FEATURE # enable use of this feature --disable-FEATURE # disable use of this feature Here is the list of FEATURES currently supported. Defaults are shown in brackets, though a value shown as `yes' (equivalent to --enable-FEATURE) will be ignored if your OS doesn't support that feature. zsh-debug # compile debugging features into zsh [no] zsh-mem # use zsh's memory allocators [no] zsh-mem-debug # debug zsh's memory allocators [no] zsh-mem-warning # turn on warnings of memory allocation errors [no] zsh-secure-free # turn on memory checking of free() [no] zsh-hash-debug # turn on debugging of internal hash tables [no] etcdir=directory # default directory for global zsh scripts [/etc] zshenv=pathname # the path to the global zshenv script [/etc/zshenv] zshrc=pathname # the path to the global zshrc script [/etc/zshrc] zlogin=pathname # the path to the global zlogin script [/etc/zlogin] zprofile=pathname # the path to the global zprofile script [/etc/zprofile] zlogout=pathname # the path to the global zlogout script [/etc/zlogout] fndir=directory # the directory where shell functions will go # [DATADIR/zsh/VERSION/functions] site-fndir=directory # the directory where site-specific functions can go # [DATADIR/zsh/site-functions] additional-path # add directories to default function path [] function-subdirs # if functions will be installed into subdirectories [no] dynamic # allow dynamically loaded binary modules [yes] largefile # allow configure check for large files [yes] locale # allow use of locale library [yes]