Principles and standards
Alongside the conceptual and content planning for the website, principles and technical standards were formulated to guide the next phase. After researching best practice for flexibility of web content, the team decided on the following principles:
- separation of content from presentation
- support for bilingual content (Māori and English) and a bilingual interface
- use of open standards
- management of content outside proprietary systems.
Site content would be prepared and delivered in such a way that the site’s look and feel could be changed and updated without having to re-create it completely. The web technology for ensuring this was CSS (Cascading Style Sheets), in conjunction with HTML encoding of the text. Applying open standards would ensure that the content could not be locked into any particular software and could be re-used in a variety of ways.
Responses to the tender process
These principles were written into the tender documentation as mandatory. This caused some head-scratching amongst the respondents, and some creative solutions. Some were more robust than others.
At one end of the spectrum were proposals to use established content management systems, designed to handle large volumes of data, customisable to a limited extent, and extremely expensive to license and implement – beyond the level of Te Ara’s resources. At the other end were purpose-built applications, promoted with enthusiasm and confidence by geeky programmers. The high end seemed likely to threaten the site with inflexible methodologies, and at the creative end there were risks to the site’s schema and intentions.
The chosen solution
A consultant project manager was employed on the tender and evaluation processes for several months during 2002. A rigorous evaluation process, carried out by Jock Phillips and Ross Somerville with the assistance of Jamie Mackay and Ian Kemp, included interviews with lead developers and reference checks.
It was decided to use the Microsoft Content Management Server (MCMS) to deliver the web pages, but with authoritative versions of the content managed outside the content management system (CMS). A ‘batch upload’ process was used to create entries on the website, using XML (eXtensible Markup Language) documents to define all elements of the web pages, including styles.
The batch upload process tied together the non-text content (images, sound files and so on, held in the ministry’s file system, with metadata managed in a relational database) and the XML documents.
It was decided to store all text in a software-neutral format. This made it easy to check that the web pages were valid, so the website would not ‘break’. The content had a defined structure for all pages and page elements, which in principle were infinitely variable and extensible, future-proofing the site.
Te Ara’s editors managed the essential work of pre-production, formatting the XML and other documents, uploading them to a staging website and publishing them to the live site. They also made and managed all updates, corrections and changes to the site content, while the design team created the digital assets, many of them from scratch, and made improvements to the way the site worked, in conjunction with contracted developers.
Not so Flash
In the early years of Te Ara, animations and interactives were built in Shockwave (later Adobe) Flash. This was the most popular program at the time, but these resources were inaccessible to screen readers and did not work on iPhones and tablets. It was decided to rebuild them all in HTML 5.
Migration to Drupal and subsequent developments
In 2009, as Microsoft had stopped supporting MCMS, and in the light of the increasing development of more flexible, open-source and non-proprietary solutions, Te Ara commissioned evaluations of a range of options. The NZHistory site had already decided to use Drupal open-source content management software, and Te Ara decided to follow suit. This would also give the Te Ara team greater control and flexibility in the evolving design of the site, and a newer, flexible technology at a lower cost.
The migration was contracted to Headfirst, who worked with Shift to convert the site’s look and feel to work with the new back end.
Scraping the site
The content migration to Drupal was complex; there was no way to programmatically extract the content from the CMS, where it was stored as BLOBs (Binary Large Objects), which could not be decoded. The site was ‘scraped’ and re-created using Drupal.
A modified version of the original XML upload process was implemented. This was simplified, as a larger amount of the file management could be done via the Drupal interface.
The site was redesigned in-house and relaunched in 2011 with improved usability and performance. In 2013 the version of Drupal was upgraded.
In 2017 the site was maintained by in-house developers with some service agreements with outside firms. There had been some convergence in the ministry’s website platforms, which promised better integration between sites and production efficiencies in the future.
