By: Sam Hames

This past year I’ve been studying part time for a Bachelor of Science (Honours) degree specialising in Computer Science at Flinders University. The focus of my studies has been my honours thesis. My thesis explored the following research question:

Is it possible to provide collaborative mapping services on mobile devices in an infrastructure independent manner?

To explore this question I developed an Android based application that could support four core objectives. They were:

1. have the users own geographic location displayed on a map;
2. add incidents, represented by a marker, onto a map;
3. be able to see the geographic location of other users of the application on the map; and
4. share details of incidents with other users of the application on the network.

The software that I developed uses the resilient Ad Hoc mesh network provided by the Serval Project to ensure that communication between instances of the application are infrastructure independent.

The thesis is available online on my Stuff by Techxplorer website.

The photo “Writing Thesis” was uploaded to Flickr by Sam Hames and used under the terms of a Creative Commons License.

By: depone

Today I released the first version of Techxplorer’s CMS. The source code is available on GitHub.

Techxplorer’s CMS is a series of PHP scripts, and accompanying JavaScript files, that I use in conjunction with MultiMarkdown to manage the Stuff by Techxplorer website. The CMS was designed with similar goals to that of the MultiMarkdown CMS. I decided to develop my own system for three main reasons:

1. I wanted to see if I could do it
2. I wanted something in PHP as I’m allergic to Perl
3. I didn’t need all of the functionality that the MultiMarkdown CMS provides

The current version of the system provides the following capabilities:

1. Convert MultiMarkdown source files into HTML files when the md file is newer or there is no corresponding HTML file
2. Match the URL of the request to a HTML file in the content directory
3. Incorporate the HTML from the file into the website template
4. Automatically add a table of contents via JavaScript when required
5. Automatically create links from citations to a bibliography when required
6. Add an image cycle to a page
7. Automatically include page, directory and site specific metadata

The photo “Content Management” was uploaded to Flickr by depone and used under the terms of a Creative Commons License.

In the morning members of the school are invited to attend and check out what we’ve been doing. Some are assigned individual students and they must assess their posters. At the same time a group of industry representatives is also working their way through the expo evaluating a number of projects for awards.

My poster, and presentation, was selected for the “Most Outstanding Software Engineering Project 2011″. On Christmas Eve the award arrived in the mail, which was a very pleasant surprise.

The poster is available as a PDF via my bytechxplorer.com site and my honours thesis is also available.

Many thanks to my supervisors for providing feedback on my poster and for giving me the opportunity to work on the Serval Project.

By: Toni Verdú Carbó

This post is a continuation of the series of posts that I’ve written on my experiences with the Lithium framework.

In the MARQues project I wanted to have a very simple way of logging the occurrence of various activities. The two activities I’m most interested in are the two types of searching that a user can undertake. I want the system to be able log when a search is undertaken and what search terms were used.

Initially I’d developed my simple activity logging table in the MySQL database with a timestamp field. Unfortunately this field doesn’t currently work with the Lithium framework. No matter what I did the field was always populated with zeros. This is a known issue and is currently marked as an enhancement.

To resolve the issue I changed the field to an ordinary datetime field and used the PHP date function to populate it. For example I use the model like this:

// build an array of values
$log = array(
  'type'  => 'search',
  'notes' => $this->request->data['search'],
  'timestamp' => date('Y-m-d H:i:s')
);

// create an instance of the model using the values
$activity = ActivityLogs::create($log);

// save the model
$activity->save();

I hope this post proves useful to other who may be experiencing similar issues.

The photo “The passage of Time” was uploaded to Flickr by Toni Verdú Carbó and used under the terms of a Creative Commons License.

By: Sarah Horrigan

Recently I discovered that the main page of a website that I’m developing wasn’t rendering properly in Internet Explorer 9 (IE9). The page in question includes HTML5 specific elements, included the appropriate doctype and the response headers included the appropriate mime-type as well.

In browsers such as Google Chrome and Mozilla Firefox the page would render as expected. In IE9 the page would render, and the icon for compatibility mode would appear. The intention of this mode, as far as I can tell, is to alert the user that the page may be using content that works best in another rendering mode. Unfortunately for me in my experience this mode doesn’t understand HTML5 specific tags. If a user used this mode the page would fail to render properly.

To fix this it is necessary to define a document compatibility mode by including an Internet Explorer specific meta tag in the head of the page.

In my case the header went from this:

<!DOCTYPE html>
<html lang="en">
  <head>
    <title>My HTML page</title>
  </head>
</html>

to this:

<!DOCTYPE html>
<html lang="en">
  <head>
    <!-- Explicitly Enable IE9 Standards mode -->
    <meta http-equiv="X-UA-Compatible" content="IE=9" >
    <title>My HTML page</title>
  </head>
</html>

There are two things that irk me about this situation. First, while it is possible to determine that IE 9 has decided that compatibility mode should be presented as an option to the user, it isn’t possible to determine why. Which makes it near impossible to diagnose what is going on.

Second, is that my standards compliant page, which validated fine according to the W3C Validator, now must contain a vendor specific extension. I was naive enough to think that the days of vendor specific extension in the HTML were going to be a thing of the past.

The photo “Little Monkey” was uploaded to Flickr by Sarah Horrigan and used under the terms of a Creative Commons license.

By: vlasta2

In an earlier post I wrote about my explorations of using MultiMarkdown to create a LaTeX document that included a bibliography. Since that post I’ve made some changes.

First, I’m now using the MultiMarkdown Composer application to author my document. I’ve only had it a few days, and so far like it very much.

Second, I’ve written a rudimentary bash script to automate the creation of the PDF document. In my current workflow I keep my source file separate from the output. Mainly because the creation of the PDF from the LaTeX file creates more than 10 additional files and I didn’t want these mixed up with my source files.

The script is as follows:

#!/bin/bash
# script to help in making the document
# set some options
set -o nounset
set -o errexit
# get the current directory
WORK_DIR=`pwd`
MMD_UTILS_DIR=/path/to the/MultiMarkdown/Utilities

# clean up the output directory
rm -f "$WORK_DIR"/output/*

# merge the source files together
"$MMD_UTILS_DIR"/mmd_merge.pl "$WORK_DIR"/index.txt > "$WORK_DIR"/output/latest.md

# convert to latex
/usr/local/bin/multimarkdown "$WORK_DIR"/output/latest.md -t latex -o "$WORK_DIR"/output/latest.tex

# change to the output directory
cd "$WORK_DIR"/output

# run pdflatex
/usr/texbin/pdflatex -halt-on-error latest.tex &> stage-01.log

# run pdflatex
/usr/texbin/pdflatex -halt-on-error latest.tex &> stage-02.log

# make the bibliography
/usr/texbin/bibtex latest.aux &> stage-03.log

# run pdflatex
/usr/texbin/pdflatex -halt-on-error latest.tex &> stage-04.log

# run pdflatex
/usr/texbin/pdflatex -halt-on-error latest.tex &> stage-05.log

The script has been developed to work on my MacBook running the lastest version Mac OS X and that also has the latest MacTeX version installed as well as the MMD Support Files and the MMD LaTeX Support Files.

The “Type ornaments” photo was uploaded to Flickr by vlasta2 and used under the terms of a Creative Commons License.

By: Alan Berning

I’ve been writing a series of posts lately on my experiences with the Lithium framework. Mainly about generating JSON as that’s what I’m working on at the moment.

This morning I noticed an issue I was having with the routes to my controller. I don’t know if it is a bug, something I don’t understand and am implementing wrong.

What I noticed was that a URL like the one below was causing output to be in JSON format, but the id number (3) at the end of the URL wasn’t being passed into my function.

http://example.com/mycontroller/myfunction/3.json

The function definition in the controller looks like this:

public function myfunction($id = null) {
	// do something with the id variable
}

The routes that I believe are allowing me to specify the content type as part of the URL is like this:

Router::connect('/{:controller}/{:action}/{:id:[0-9]+}.{:type}', array('id' => null));
Router::connect('/{:controller}/{:action}.{:type}');

For the moment I’ve needed to adjust my function so that it uses the Request object associated with the controller when the function runs to look for the id parameter manually if the request is for JSON data like this:

public function myfunction($id = null) {
	// check if this is a request for JSON data
	// and if so retrieve the id value
	if($id == null && $this->request->type() == 'json') {
		$id = $this->request->params['id'];
	}
}

If anyone else is having issues like this I hope this post helps, and if anyone knows what I’m doing wrong please leave a comment.

The “route 66” photo was uploaded to Flickr by Alan Berning and used under the terms of a Creative Commons license.

By: EMSL

Last Friday I wrote a post about my initial explorations in revisiting the topic of using a Lithium controller to output JSON encoded data. In the post I mentioned a technique that I’d determined would allow me to check if a request to a controller was for JSON encoded data or not.

The code looked like this:

if($this->request->type() == 'json') {
	// build an array of elements and return it
} else {
	// build data elements to be used in the template
}

Over the weekend I determined a way to provide an enhanced experience to the user using this technique.

In essence what my code now does is output a basic page when a request that should be for JSON data is not constructed correctly i.e. when it is missing the .json suffix.

As an example my code now looks like this:

if($this->request->type() == 'json') {
	// build an array of elements and return it
} else {
	// build data elements to be used in the template
	// get the data

	// set a title
	$title = 'Did you mean to get JSON data?';

	// get the URL used for the request
	$url = $this->request->url;

	// use a basic layout
	$this->_render['layout'] = 'not_json';

	// build the page
	return compact('data', 'title', 'url');
}

In essence:

• Line 05 is where you would retrieve the data that should be returned in JSON format, or at least a subset that makes the page more useful
• Line 08 is a descriptive title for the page
• Line 11 is the URL that was used to request the page
• Line 14 changes the layout used to build the page from the default to one called not_json
• Line 17 returns this data and renders the page

The key here is that this basic structure could be extended to provide additional information and not just a limited HTML page that illustrates that an error was made.

The not_json layout is a file (not_json.html.php) that lives in the app/views/layouts directory and is a very basic HTML page. It could be a more complex page, for example one that responds with an error code or contains additional explanatory information. For my purposes a very basic HTML page will suffice for the time being.

The “Lithium Manganese Phosphate” photo was uploaded to Flickr by EMSL and used under the terms of a Creative Commons license.

The key to determining if the request was for JSON data is the Request class, specifically the type function.

As an example the code in my controller looks like this:

if($this->request->type() == 'json') {
	// build an array of elements and return it
} else {
	// build data elements to be used in the template
}

In this way a request for example http://example.com/example.json will return data in a JSON encoded format. A request for http://example.com/example will return the data in HTML, some other format, or even produce an error.

By: Shin

One of the more critical components of any open source system is the documentation. In a previous project the team I was a part of was complimented on our documentation and how we kept it up to date. One of the hardest aspects of a system to document, and keep up to date, is the database schema. With some prior thought it can be made easier.

The first aspect of the database to document is to add comments where appropriate. As an example in the MARQues system I’ve added a brief comment to each field and table. These comments are available in GUI client that I use and also forms part of the database-schema file.

The second aspect is to define the foreign key relationships between tables. In MySQL this means I use the InnoDB table type where possible.

The third aspect is to use the database-schema file to create a diagram of the database structure. For this I use the mysqlviz script which uses the database-schema file to create a script in the DOT language. This script is then used by the Graphviz application to create the final image.

This sounds like a lot of work but I’ve written a small script which takes care of all of the steps which looks like this:

#!/bin/bash
# a script to make it easier to export the database schema
# and create a basic schema diagram
# set some variables
# adjust accordingly to match your own environment
CURR_PATH=`pwd`
OUTPUT_FILE=database-schema.sql
IMAGE_FILE=database-schema.png
MYSQLDUMP_CMD=/opt/local/bin/mysqldump5
DOT_CMD=/usr/local/bin/dot
USER=mysql_user
PASSWORD=mysql_password
DATABASE=mysql_database
# do the export
$MYSQLDUMP_CMD --user=$USER --password=$PASSWORD --no-data --comments --dump-date $DATABASE > $CURR_PATH/$OUTPUT_FILE
# turn the dump into an image
$CURR_PATH/mysqlviz -f $CURR_PATH/$OUTPUT_FILE | $DOT_CMD -Tpng > $CURR_PATH/$IMAGE_FILE

Obviously I’m using MySQL as my database platform which I’ve installed using the technique I posted about yesterday. You can see an example of the output here.

The last aspect is to include additional text where necessary. For the MARQues system I keep this page in the wiki up to date with explanatory text about the database.

The “Original Database” photo was uploaded to Flickr by Shin and used under the terms of a Creative Commons license.