Content from 2016-04
posted on 2016-04-22 by Daniel "jackdaniel" Kochmański
In this short tutorial I'll describe how to bootstrap easily a project website. In fact that's what I did today with the Embeddable Common-Lisp website in order to provide the RSS feed and make putting there the news easier.
Additionally I'm showing here, how to create a standalone executable
clon after providing
bundle of systems.
First clone the repository:
$ cd /home/p/ecl $ git clone https://gitlab.common-lisp.net/dkochmanski/sclp.git website $ cd website
Now you should adjust the appropriate files. Edit
is self-explanatory), static pages and posts.
Each file with the extension
*.page is a static
pages/main.page is an example template with a static page –
don't forget to link it in the
section. Exact URL of the page is declared in the file's header.
*.post represent blog/news posts which appear in the RSS
feed. They are indexed and accessible from the root URL. Supported
file formats are
cl-who (if enabled).
When you're done, you could just load coleslaw with your favorite CL
implementation, using Quicklisp load
coleslaw and call the function
main on the website directory:
(ql:quickload 'coleslaw) (coleslaw:main "/home/p/ecl/website/")
We will take more ambitious road – we'll create a standalone executable with a proper command line arguments built from a clean bundle produced by Zach Beane's Quicklisp. CLI arguments will be handled by Clon – the Command-Line Options Nuker, an excellent deployment solution created by Didier Verna.
Creating the bundle
Bundle is a self-containing tree of systems packed with their
dependencies. It doesn't require internet access or
Quicklisp and is
a preferred solution for the application deployment.
Some dependencies aren't correctly detected –
possibly know, that our plugin will depend on the
cl-who system, and
it can't detect
cl-unicode's requirement during the build phase –
flexi-streams (this is probably a bug). We have to mention these
Clon is added to enable the clonification (keep reading).
(ql:bundle-systems '(coleslaw flexi-streams cl-who cl-fad net.didierverna.clon) :to #P"/tmp/clw")
Clonifying the application
(in-package :cl-user) (require "asdf") (load "bundle") (asdf:load-system :net.didierverna.clon) (asdf:load-system :coleslaw) (asdf:load-system :cl-fad) (use-package :net.didierverna.clon) (defsynopsis (:postfix "DIR*") (text :contents "Application builds websites from provided directories.") (flag :short-name "h" :long-name "help" :description "Print this help and exit.")) (defun main () "Entry point for our standalone application." (make-context) (when (getopt :short-name "h") (help) (exit)) (print (remainder)) (handler-case (mapcar #'(lambda (p) (coleslaw:main (cl-fad:pathname-as-directory p))) (remainder)) (error (c) (format t "Generating website failed:~%~A" c))) (terpri) (exit)) (dump "coleslaw" main)
You may generate the executable with
ecl has some
problems with the
coleslaw dependency -
esrap, I'm working on
it). I have used
ccl, because it doesn't "derp" on the symbol
and produces slighly smaller executable than
Issue the following in the bundle directory (
ccl -n -l clonify.lisp
This command should create native executable named
coleslaw in the
same directory. On my host
ccl produces binary with the approximate
This is a very simple executable definition. You may extend it with new arguments, more elaborate help messages, even colors.
To generate a websites with sources in directories
/tmp/b you call it as follows:
./coleslaw /tmp/a /tmp/b
That's all. Deployment destination is set in the
.coleslawrc file in
each website directory.
Adding GIT hooks
You may configure a post-receive hook for your GIT repository, so your
website will be automatically regenerated on each commit. Let's
assume, that you have put the
coleslaw standalone executable in
place accessible with the
PATH environment variable. Enter your bare
git repository and create the file
cd website.git cat > hooks/post-receive <<EOF ########## CONFIGURATION VALUES ########## TMP_GIT_CLONE=$HOME/tmp-my-website/ ########## DON'T EDIT ANYTHING BELOW THIS LINE ########## if cd `dirname "$0"`/..; then GIT_REPO=`pwd` cd $OLDPWD || exit 1 else exit 1 fi git clone $GIT_REPO $TMP_GIT_CLONE || exit 1 while read oldrev newrev refname; do if [ $refname = "refs/heads/master" ]; then echo -e "\n Master updated. Running coleslaw...\n" coleslaw $TMP_GIT_CLONE fi done rm -rf $TMP_GIT_CLONE exit EOF
That's all. Now, when you push to the master branch your website will
be regenerated. By default
.gitignore file lists directory
static/files as ignored to avoid keeping binary files in the
repository. If you copy something to the static directory you will
have to run coleslaw by hand.
Coleslaw is a very nice project simplifying managing project website
with easy bootstrapping the site without any need to maintain working
lisp process on the server (this is static content which may be served
apache) and allowing easy blogging (write a post in
markdown and push to the repository).
Sample Common-Lisp Project is a pre-configured website definition
with a theme inspired by the
common-lisp.net projects themes with
some nice features, like RSS feed and blog engine (thanks to
We have described the process of creating a simple website, creating a standalone executable (which may be shared by various users) and chaining it with git hooks.