Sun Feb 22 09:08:47 1998, faith@cs.unc.edu
WHAT:
The files contained herein are designed to convert the data from the
ftp://ftp.cs.unc.edu/pub/users/faith/dict/web1913-0.46-a.tar.gz
package into a working database and index for the DICT server.
The original data for this version of:
Webster's Revised Unabridged Dictionary (G & C. Merriam Co., 1913,
edited by Noah Porter). Online version prepared by MICRA, Inc.,
Plainfield, N.J. and edited by Patrick Cassidy <cassidy@micra.com>.
ftp://uiarchive.cso.uiuc.edu/pub/etext/gutenberg/etext96/pgw*
http://humanities.uchicago.edu/forms_unrest/webster.form.html
is available from: ftp://ftp.uga.edu/pub/misc/webster/, but some patches
were applied to the files in the web1913-0.46-a.tar.gz distribution.
This package only contains conversion software, and contains NO DATA. You
must obtain the data separately. Please see the raw data for distribution
terms for the raw and formatted data. The formatting _software_ is
currently distributed under the GNU General Public License. If you need it
use it under another license, contact the author.
CONFIGURATION:
./configure options:
--with-tmppath=PATH use PATH for temporary files
--with-datapath=PATH use PATH for original database files
--prefix=PREFIX install architecture-independent files in PREFIX
--datadir=DIR read-only architecture-independent data in DIR
tmppath defaults to . and requires 40MB
datapath defaults to ./web1913-* and should contain the data found in
web1913-0.46-a.tar.gz [this software probably won't work with later
versions of the raw data, but you can try...]
prefix defaults to /usr
datadir defaults to /usr/lib, so the data files get put in /usr/lib/dict.
They are machine independent, you you'd want to change this to a shared
directory structure if you want the data shared between several servers.
For example, use --datadir=/usr/share to put the files in /usr/share/dict.
30 MB of data and 4MB of index will be written to ./web1913.dict and
./web1913.index -- you must have enough disk space available for these
files, for the original 42MB of raw files (in datapath), and for the
intermediate 40MB file (in tmppath). This means you need close to 150MB of
free disk space to format this dataset. Additionally, you need about 50MB
of swap space for the running programs. Formatting takes less than 15
minutes on a typical Pentium 133MHz.
If you have libmaa installed on your system, then the local copy won't get
built unless you use --with-local-libmaa. Otherwise, it will be built
automatically.
If you have the dictzip program installed on your system, then the final
30MB web1913.dict file will be compressed to about 10MB. The file will be
readable with the GNU zip utilities. If you don't want this compression
performed even though you have dictzip installed, make DICTZIP=cat and it
will be skipped.
EXECUTION:
make builds the libraries and executables
make db formats the database
make install installs the database (no executables are installed)
When you're done with the "make db" step, it should say
185399 headwords words for 115301 entries, total
[Versus 185774 headwords words for 115272 entries, total in 0.45.]
[Versus 187637 headwords words for 115283 entries, total in 0.42]
[Versus 202209 headwords words for 124121 entries, total in cide 0.43]
and you'll be using up about 80MB for everything (less if you have dictzip
installed and it automatically compressed the database). Note that each of
the processes use about 40MB of memory, so you need a reasonable amount of
RAM or swap space.
PROBLEMS:
If the problem is with the data, you should contact MICRA via Patrick
Cassidy <cassidy@micra.com>. Be nice and try to be helpful -- putting this
database together was a lot of work, and there are a lot of typos and
data-related problems. If you want to edit definitions, you should
probably get the latest version of the data and work with that. If you
want to redistribute the data, see the terms in the raw and formatted data
files (the terms are MICRA's, and are the same for the raw and formatted
data).
If the problem is with the programs, you should contact me, Rik Faith
(faith@cs.unc.edu). Patches or suggestions are always welcome, but my time
is limited and I might not reply to your email immediately. If its many
years after 1998, faith@acm.org might be a better alternative address.