Diff for /gforth/INSTALL between versions 1.5 and 1.25

version 1.5, 1995/01/21 17:26:40 version 1.25, 1996/11/07 22:31:31
Line 1 Line 1
 Preliminary version  
   
 You need gcc version 2.0 or later to compile gforth.  You need gcc version 2.0 or later to compile gforth.
   
 First, type  First, type
   
 configure  ./configure
   
   (see Section Configuration Options below for details).
   
   After configuration, type
   
   make
   
   Now you can check whether your shiny new Forth system works. Say
   
   make test
   
   You can run some benchmarks with
   
   make bench
   
   and compare them with the results in Benchres and in the manual.
   
   If everything is all right, you may want to install gforth. Type
   
   make install
   
   You have to make an entry in the info directory file manually.
   
   For paper documentation, print gforth.ps (a Postscript file (300dpi
   fonts, i.e., it works, but does not produce best quality on better
   printers)), or say
   
   make gforth.dvi
   
   and print the resulting file gforth.dvi. You can also get the
   documentation in HTML format by typing
   
   make html
   
   If you prefer plain ASCII documentation, just concatenate the files
   gforth.info-* ('cat gforth.info-*' under Unix).
   
   
                   Configuration Options
   
   If you use GNU make, you can build in a directory different from the
   source directory by changing to the build directory and invoking
   configure thus:
   
   $srcdir/configure
   
   where $srcdir is the source directory. (Note that we tested this only
   for installation; i.e., if you want to hack the Gforth sources, you
   should probably build in the source directory).
   
 configure has the following useful parameters:  configure has the following useful parameters:
 -direct-threaded: setup for a direct threaded interpreter; this is faster    --prefix=PREFIX         install architecture-independent files in PREFIX
    on many machines, but needs special support. Unsupported machines will                            [default: /usr/local]
    ingore this switch. On some CISC machines, direct threading isn't an    --exec-prefix=PREFIX    install architecture-dependent files in PREFIX
    advantage over indirect threading.                            [default: same as prefix]
     --enable-force-reg      Use explicit register declarations if they appear in
                             the machine.h file. This can cause a good speedup,
                             but also incorrect code with some gcc versions on
                             some processors (default disabled).
     --enable-direct-threaded      Force direct threading. This may not work on
                                   some machines and may cause slowdown on others.
                                   (default processor-dependent)
     --enable-indirect-threaded    Force indirect threading. This can cause a
                                   slowdown on some machines.
                                   (default processor-dependent)
     --with-debug     specifies option -g to compile with debug info (default)
     --without-debug  omits the -g switch and creates smaller images on
                      machines where strip has problems with gcc style
                      debugging informations.
     --help: tells you about other parameters.
   
 -without-debug: omits the -g switch and creates smaller images on machines  The file Benchres shows which combination of the -enable options we
    where "strip" has problems with gcc style debugging informations.  tried gave the best results for various machines.
   
 --help: tells you about other parameters.  If you don't like the defaults for the installation directories, you
   should override them already during configure.  E.g., if you want to
   install in the /gnu hierarchy instead of in the default /usr/local
   hierarchy, say
   
 Now type  ./configure --prefix=/gnu
   
 make  Moreover, if your GCC is not called gcc (but, e.g., gcc-2.7.1), you
   should say so during configuration. E.g.:
   
   env CC=gcc-2.7.1 ./configure
   
   You can also pass additional options to gcc in this way, e.g., if you
   want to generate an a.out executable under Linux with gcc-2.7.0:
   
   env "CC=gcc -b i486-linuxaout -V 2.7.0" ./configure
   
   You can change the sizes of the various areas used in the default
   image `gforth.fi' by passing the appropriate Gforth command line
   options in the FORTHSIZES environment variable:
   
   env "FORTHSIZES=--dictionary-size=256k --data-stack-size=16k --fp-stack-size=16k --return-stack-size=16k --locals-stack-size=16k" ./configure
   
   The line above reaffirms the default sizes. Note that the locals
   stack area is also used as input buffer stack.
   
   If C's "long long" do not work properly on your machine (i.e., if the
   tests involving double-cell numbers fail), you can build Gforth such
   that it does not use "long long":
   
   env ac_cv_sizeof_long_long=0 ./configure
   
   
                           Cross-Configuration
   
   A few tests made by the configure script do not work in a
   cross-compilation situation. You have to provide the results of these
   tests by hand. E.g., if you compile for a 386 architecture processor:
   
   env ac_cv_sizeof_char_p=4 ac_cv_sizeof_short=2 ac_cv_sizeof_int=4 ac_cv_sizeof_long=4 ac_cv_sizeof_long_long=8 ac_cv_c_bigendian=no ./configure
   
   The ac_cv_sizeof_... variables give the sizes of various C types;
   ac_cv_sizeof_char_p is the same as "sizeof(char*)" in C code. The
   ac_cv_c_bigendian variable gives the byte order.
   
   
                   Preloading installation-specific code
   
   If you want to have some installation-specific files loaded when
   Gforth starts (e.g., an assembler for your processor), put commands
   for loading them into /usr/local/share/gforth/site-forth/site-init.fs
   (if the commands work for all architectures) or
   /usr/local/lib/gforth/site-forth/site-init.fs (for
   architecture-specific commands);
   /usr/local/lib/gforth/site-forth/site-init.fs takes precedence if both
   files are present (unless you change the search path). The file names
   given above are the defaults; if you have changed the prefix, you have
   to replace "/usr/local" in these names with your prefix.
   
   By default, the installation procedure creates an empty
   /usr/local/share/gforth/site-forth/site-init.fs if there is no such
   file.
   
   If you change the site-init.fs file, you should run "make install"
   again for the changes to take effect (Actually, the part of "make
   install" starting with "rm gforth.fi" is sufficient).
   
   
                   Multiple Versions and Deinstallation
   
   Several versions of Gforth can be installed and used at the same
   time. Version `foo' can be invoked with `gforth-foo'. We recommend to
   keep the old version for some time after a new one has been installed.
   
   You can deinstall this version of Gforth with 'make uninstall' and
   version foo with 'make uninstall VERSION=foo'. 'make uninstall' also
   tells you how to uninstall Gforth completely.
   
   
                           A Possible Problem
   
   You need to read this only if you see a message like
   
   ...
   The Gforth installer should look into the INSTALL file
   
   1) "gforth: Cannot load nonrelocatable image (compiled for address $1234) at address $5678
   The Gforth installer should look into the INSTALL file"
   
   Gforth supports both relocatable and fixed-address images. If you load
   normal Forth code and save the image, you get a fixed-address
   image. Producing a relocatable image is more difficult.
   
   Therefore, Gforth has only a relocatable image of the kernel
   (kernel.fi), which is powerful enough to load the rest of
   Gforth. However, loading the rest takes a noticable amount of time. To
   avoid this delay (which would occur on every startup), the
   installation procedure produces an image fixed at an address
   determined at the Gforth run that produced the image. This
   fixed-address image is loaded by default. On most OSs this works,
   because the first chunk of memory is always allocated at the same
   address. If the address changes, you get the message above.
   
   An image address change can be caused by a change of the gforth
   executable, or by a change (upgrade) of the OS; in these cases you
   just have to rebuild and reinstall the fixed address image with
   
   rm gforth.fi; make gforth.fi; make install
   
   If you get such a message with a different address in place of the
   $5678 each time you try to start gforth, you cannot use fixed-address
   images on your OS. In this case, send us a message so that we start
   searching for a comfortable solution to this problem. In the
   meantime, start gforth with
   
   gforth -i kernel.fi startup.fs
   
   If the addresses changes by only a small amount (e.g. by one or two
   pages), you can fix it by defining FUZZ (in config.h) to a number at
   least two times the changes you observe (0x4000 is a good idea, this
   is four 4k pages) and recompile. We do this for the DJGPP port for
   DOS, because the start address there changes by one or two pages, and
   it helps us to keep the DOS people happy without investing too much
   work in a braindead environment.
   
   
   2) "%s: Checksum of image ($13579b) does not match the executable ($2468a)
   The Gforth installer should look into the INSTALL file"
   
   A fixed-address image is not only fixed with respect to its base
   address, but also with respect to certain addresses in the gforth
   executable and the threading method. These things are encoded in a
   checksum.
   
   If the checksum of the executable and the checksum of the image are
   not equal, you get the message above. This can be caused, e.g., by
   trying to run an image produced for a direct threading system on an
   indirect threaded system.
   
 If your make has trouble with the Makefile, "make gforth" might work.  Chances are that you unintentionally tried to execute an image from
   the wrong directory. As a remedy, you can specify Gforth's search
   path with the "-p" command line option and with the GFORTHPATH
   environment variable.
   
 If your installed gcc isn't called "gcc" (eg. called "gcc-2.6.1"), type  On the other hand, if you need to solve the problem by creating a new
   fixed-address image, you can use the steps described above.
   
 make CC=<whatever you call your gcc>  
   
 instead.  

Removed from v.1.5  
changed lines
  Added in v.1.25


FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>