There are three ways to use SWISH, which we list in increasing order of complexity:
SWISH can be used to access SWI-Prolog at the address below. We try to keep this server continuously online. You can use these servers for playing, courses or sharing and discussing ideas.
- https://swish.swi-prolog.org/ (plain Prolog and R)
- http://cplint.ml.unife.it/ (probabilistic and machine learning extensions)
- http://lpsdemo.interprolog.com/example/FirstStepswithLPS.swinb (Logic Production Systems)
We have not yet dealt with scalable hosting nor with really reliable and scalable storage for saved programs. We hope to keep all your programs online for at least multiple years.
We maintain Docker images at the swipl organization at Docker Hub. A bluffer's guide to run SWISH with R if you have Docker installed is as simple as this:
docker run -d --net=none --name=rserve swipl/rserve
docker run -d -p 3050:3050 --volumes-from rserve -v $(pwd):/data swipl/swish
There are many configuration options for SWISH, notably for authentication, email notifications and extension plugins. See the docker-swish repo for details.
cd
to your swish root directory and run
git submodule update --init
If you have make
installed you can configure the desired packs by
editing the PACKS
variable and run the following to download them and
configure those that need to be configured.
make packs
Install Yarn for your platform. On Ubuntu, this
implies getting node
and npm
by installing two packages and next use
npm
to install yarn
(some older Linux versions need nodejs-legacy
instead of nodejs
):
sudo apt install npm nodejs
sudo npm i yarn -g
Once you have yarn
, run the following from the toplevel of swish
to
get the dependencies:
yarn
make src
As installing node and yarn is not a pleasure on all operating systems,
you can also download the dependencies as a single zip file from
http://www.swi-prolog.org/download/swish/swish-node-modules.zip.
Unpack the zip file, maintaining the directory structure, from the swish
root directory to create the directory web/node_modules. If you have
make
installed you can install the above .zip
file using
make yarn-zip
Last updated: Dec 16, 2019: upgraded dependencies, new archive name
Install the latest SWI-Prolog development version. As SWISH is very much in flux and depends on the recent SWI-Prolog pengines and sandboxing libraries, it is quite common that you need the nightly build (Windows) or build the system from the current git development repository swipl-devel.git.
Apr 25, 2019: Works for a quite large range of SWI-Prolog versions. The current swipl-devel.git snapshot fixes an issue in CSV downloading, emitting CORS and cache control HTTP headers near the end of the CSV output.
Rendering Prolog terms as
graphs
requires Graphviz. The avatar system
requires the convert
utility from
ImageMagic. These are available as
packages for virtually any Linux system, e.g., on Debian based systems
do
sudo apt-get install imagemagick
sudo apt-get install graphviz
With a sufficiently recent Prolog installed, start the system by opening
run.pl
either by running swipl run.pl
(Unix) or opening run.pl
from the Windows explorer.
Now direct your browser to http://localhost:3050/
If you want to know what the latest version looks like, go to https://swish.swi-prolog.org/
There is a lot that can be configured in SWISH. Roughly:
-
Make additional libraries available, e.g., RDF support, database connections, link to R, etc.
-
Configure authentication and authorization. The default is not to demand and run commands sandboxed. At the other extreme you can configure the system to demand login for all access and provide full access to Prolog.
Configuration is done by reading *.pl
files from the directory
config-enabled
. The directory config-available
contains templates
that can be copied and optionally edited to create a configuration.
See README.md in config-available for details.
By default, SWISH does not require the user to login but lets you run
only safe commands. If you want to use SWISH for unrestricted
development, enable the config file auth_http_always.pl
:
mkdir -p config-enabled
(cd config-enabled && ln -s ../config-available/auth_http_always.pl)
Next, for first usage, you need to create a user. The authentication
module defines swish_add_user/0, which asks for details about the user
to be created and updates or creates a file called passwd
. At the
moment Group and E-Mail are stored, but not used.
?- swish_add_user.
% Password file: /home/jan/src/prolog/swish/passwd (update)
User name: bob
Real name: Bob Hacker
Group: user
E-Mail: [email protected]
Password:
(again):
true.
If you now try to run a command in SWISH, it will prompt for a user and password. After authentication you can run any Prolog predicate.
NOTE Authentication uses HTTP digest authentication by default. This authentication method uses a challenge-response method to verify the password and ensures the credentials change with every request such that old credentials cannot be re-used by an attacker. Unfortunately, the server stores the password as the SHA1 hash created from the user, password and realm. This is relatively vulnerable to brute-force attacks for anyone who gains access to the password file due to the low computational overhead of SHA1 and the lack of a user-specific salt. Also note that the exchanged commands and replies are not encrypted. Secure servers should use HTTPS.
Instead of using auth_http_always.pl
you can use auth_http.pl
, which
allows for unauthenticated -sandboxed- usage as well as logging in to
the server and get unrestricted access. In addition, several social
login modules are provided to login using Google, etc. Currently these
provide no additional rights. A more fine grained authorization scheme
is planned.
The script daemon.pl is provided to run SWISH as a service or daemon on Unix systems. Run this to get an overview of the options.
./daemon.pl --help
This script can be used to start SWISH as a daemon from the command
line, start SWISH from service managers such as upstart
or systemd
and simplifies running as an HTTPS server. See
https://github.com/triska/letswicrypt.
You can run SWISH alongside your normal Prolog development tools. The
cleanest way to do so is by using myswish.pl
and install this in your
local Prolog library. See myswish.pl
for details on how to set this
up.
Most of the application is realised using client-side JavaScript, which
can be found in the directory web/js
. The JavaScript files use
RequireJS for dependency tracking and
jQuery for structuring the JavaScript as jQuery
plugins. The accompanying CSS is in web/css
. More details about the
organization of the JavaScript is in web/js/README.md
There are two overal pages. web/swish.html
provides a static page and
lib/page.pl
provides a Prolog frontend to generate the overal page or
parts thereof dynamically. The latter facilitates smoothless embedding
in SWI-Prolog web applications.
No building is needed to run the system from sources. For public installations you probably want to create the minified JavaScript and CSS files to reduce network traffic and startup time. You need some more tools for that:
% sudo npm install -g jsdoc
% sudo npm install -g requirejs
% sudo npm install -g clean-css-cli
You also need GNU make installed as make
and SWI-Prolog as swipl
.
With all that in place, the following command creates the minified
versions:
% make min
The default main page (/
) is generated from lib/page.pl
. It uses
minified JavaScript and CSS from web/js/swish-min.js
web/css/swish-min.css
when available. If the minified files are not
present, the server automatically includes the full source. The
generated files may be removed using
make clean
Alternatively, use of the minified files can be disable from Prolog using this command and reloading the page:
?- debug(nominified).
The JavaScript is documented using JsDoc. The
generated documentation is available in web/js/doc/index.html
.
cplint and cplint_r related documentation is availble respectively here and here