new copyright; att/bsd/shared

[unix-history] / usr / src / usr.bin / spell / README

/*-
 * Copyright (c) 1987 The Regents of the University of California.
 * All rights reserved.
 *
 * %sccs.include.proprietary.c%
 */

#ifndef lint
static char sccsid[] = "@(#)README	1.6 (Berkeley) %G%";
#endif /* not lint */

All files and subdirectories of /usr/share/dict are recommended for
rdisting except web2 and web2a (because of their size), and some of
the files  hlist*  depending on needs of your machine (details below).
Descriptions of most of these files are given under FILES below.

The new subdirectory "special" contains lists of words in specialized
fields, which may be hashed in with the regular lists on machines having
many users working in these fields.  As of this writing, there are two
such specialized wordlists.

It is advised that system managers also create a directory
/usr/local/dict. This can be used to maintain files of particular
interest to users of each machine (e.g., surnames of members of the
department on a departmental machine).

The hashed wordlists hlista and hlistb in this distribution include
the words in the file special/4bsd, comprising current 4bsd
commands, system calls, etc. (from "abs" to "zcat").  Machines
whose primary users are programmers should take these files by
rdist.  For machines with other user populations, a file "hlist" is
provided which only contains the contents of /usr/dict/words.  Managers
of such machines should rdist this file, and use "spellin" to produce
files hlist{a,b} which contain the words from
/usr/share/dict/{american,british} respectively, plus any other files
appropriate to the needs of the majority of their users.  (Some basic
unix commands and terms that general users are likely to encounter,
e.g. troff, emacs, tty, have been included in /usr/share/dict/words.
More may be added as suggestions are received.)  Here, for instance is a
script that might be used to create the hashlists on a particular
machine, so as to include the words in /usr/share/dict/special/math, as well
as two local lists which we will assume are called
/usr/local/dict/surnames and /usr/local/dict/acronyms.

	#
	cd /usr/share/dict
	cat american special/math /usr/local/dict/{surnames,acronyms} | \
		spellin hlist > hlista
	cat british  special/math /usr/local/dict/{surnames,acronyms} | \
		spellin hlist > hlistb

     Hashlists can also be created from scratch using
/usr/src/usr.bin/spell/Makefile.  This is now written so that if "make"
is run with no options it will produce the hashed files as presently
distributed, but so that the extra wordlists used can be controlled with
variables LOCAL and SPECIAL.  For instance, the results given by the
above script can be obtained by doing:

	cd /usr/src/usr.bin/spell
	make LOCAL='/usr/local/dict/surnames /usr/local/dict/acronyms' \
		SPECIAL=special.math 
	make install

     Returning to the subject of the wordlists in /usr/dict, these are,
in general, ordered as in  sort -df.  This makes no difference for
spell's hashing process, but makes a difference for other commands,
such as "look", that perform binary searches on the unhashed lists.

Complaints, and any additional suggestions for words or wordlists,
should be sent to me.  I cannot fix bugs involving the code of "spell",
but I am maintaining a list of these bugs, and of other ideas for
improvement.
		George Bergman, gbergman@cartan.Berkeley.Edu
		18 March, 1987

--------------------------------------------------------------------
FILES and subdirectories of /usr/share/dict:

    words    -- common words, and important technical terms from all
	fields, that are spelled the same in British and American usage.
    american -- spellings preferred in American but not British usage.
    british  -- spellings preferred in British but not American usage.
    stop     -- forms that would otherwise be derivable by "spell" from
	words in one of the above files, but should not be accepted.
    hlist    -- hashed list, formed from the file "words" only.
    hlista   -- hashed list, formed from files {words,american,special/4bsd}.
    hlistb   -- hashed list, formed from files {words,british,special/4bsd}.
    hstop    -- hashed list, formed from file "stop".
    web2     -- words from Webster's 2nd International (see WEB below).
    web2a    -- compounds and phrases from same source.
    README   -- this file
    papers/  -- an (out-of-date specialized) bibliographical database,
	used as the default by the program "refer".
    special/ -- directory of less common terms from specialized fields.
	It presently contains:

	special/4bsd -- commands and system calls (from filenames in
	    /usr/share/man/man[1238n]), and builtin csh commands (named in
	    /usr/share/man/man1/csh.1) of the current version of 4bsd Unix.
	    (Supersedes old "/usr/src/usr.bin/spell/local".)
	special/math -- some mathematical terms not in /usr/share/dict/words.

WEB ---- (introduction provided by jaw@riacs) -------------------------

Welcome to Webster's Second International, all 234,936 words worth.
The 1934 copyright has elapsed, according to the supplier.  The
supplemental 'web2a' list contains hyphenated terms as well as assorted
noun and adverbial phrases.  The wordlist makes a dandy 'grep' victim.

     -- James A. Woods    {ihnp4,hplabs}!ames!jaw    (or jaw@riacs)