[ previous ] [ contents ] [ next ]
Installation instructions
The following instructions are for Unix-like systems; specifically,
arfg was written and used under various versions of GNU/Linux. For
other operating systems your mileage will vary, even though arfg can
certainly made to run wherever a reasonably recent copy of Perl and
an implementation of m4 are available.
Basically, arfg assembles a pipeline of calls to a number of
filters, and calls system() on the resulting command
string. Thus, for arfg to work properly, the following tools must be
installed:
- m4 (/usr/bin/m4):
- This macro processor should already be installed on most systems;
in case it is not, you can get the GNU-version from your favorite GNU mirror, or directly
from, e. g., ftp://ftp.seindal.dk/pub/rene/gnu/m4-1.4.tar.gz;
version 1.4 is known to work with arfg, older or non-GNU versions
should probably be avoided.
- divert
(/usr/local/bin/divert):
- This diversion filter for accumulating and re-locating blocks of
text is part of Ralf S. Engelschall's Web Meta
Language (wml). Consequently, there are the following
possibilities for installing:
- eperl
(/usr/local/bin/eperl):
- The «embedded Perl interpreter» is also part of wml. Again,
there are several possibilities for installing:
- Download and install wml; then, copy or
link the file wml_p3_eperl to
/usr/local/bin/eperl.
- Download and install eperl directly
as released by Ralf S. Engelschall, its original author.
- Download and install the new re-write of eperl by David Ljung Madison. This new version,
released under a licence that does not qualify as a free software
licence, is itself written in Perl, which makes it much slower than
the compiled version but installation just means copying the file. As
of this writing, there are two different versions available from http://freshmeat.net/projects/ePerl/: the minimal version is not
fully compatible to the original written in C, but the full version, intended as a
drop-in replacement for Engelschall's implementation, is even
slower than the minimal one. Bear in mind, however, that arfg code
meant for distribution should not rely on these incompatibilities.
There have been reports on problems building the compiled versions
eperl with Perl 5.6; in these cases, a working version of eperl
may be obtained via CVS. Also, the re-implementation in
Perl should be free from such problems.
- cat (/bin/cat):
- Every Unix system should have a cat command. arfg expects to
find it in /bin/cat.
- sh-compatible shell:
- Again, on Unix this should already be installed (usually
sh itself or a compatible shell like bash). This
program is not used by arfg directly but only via Perl's
system() call.
In the above list the full paths assumed by arfg have been
indicated. If the executables' locations do not match those indicated,
link or copy them to the expected locations or edit the arfg script
(see below).
As arfg itself is just a Perl script, installation is trivial:
- Download, unpack, extract any of the following
files:
- If necessary, edit the file arfg contained therein to adjust the
paths for the executables to be used. This might also be useful when
you want arfg to have a private copy so that compatibility is not
broken when any of the tools arfg depends upon is upgraded or
changed.
- Copy or link the file arfg to a directory on your shell's search
path.
- Make sure the permissions and ownership are correct.
- The downloaded archive also includes a copy of these web pages
that you might want to save; you should, however, always be able to
access up-to-date information by pointing your browser to http://purl.oclc.org/NET/arfg/. (The URL is displayed as part of arfg's help message.)
[ previous ] [ contents ] [ next ]