Good Practice Guide for Developers of Cultural Heritage Web Services
Good Practice Guide for Developers of Cultural Heritage Web ServicesAs the use of digital networks becomes pervasive, the imperative to develop Web sites and Web services that are accessible to all users becomes apparent. Much work has been done since the inception of the Internet to understand the issues surrounding accessibility and groups such as the WAI at the World Wide Web Consortium have produced structured guidelines to assist developers.
The subject of digital accessibility is a very broad one and encompasses many different sub-sets of user requirements accordingly. The challenge facing Web site developers is to harmoniously integrate the formal requirements of the design and function of their sites with the current accessibility standards and guidelines.
Responsibility for the publication of a publicly-accessible Web site demands that the developers understand the very diverse range of means by which their online resources may be accessed and used. Satisfactory appearance and usage of a site depends upon firstly understanding the audience and their requirements then applying that understanding in the specification and development of the site.
The two areas of accessibility and usability have significant overlap in scope and intent. Put simply, a 'usable' site can be considered to be one that is fully accessible to all users whatever their means of access and physical impairment. However, usability criteria extend into less definable elements regarding process and work flow required by both the site structure and the user interface. All projects aiming to develop material for Internet delivery should have as one of their foremost goals the attainment of a fully usable service and that within that goal there should be a clear acknowledgement that this is only achieved once all accessibility issues have been satisfactorily resolved.
The authoritive group concerned with the accessibility of Web resources is Web Accessibility Initiative (WAI), a group within the umbrella of the World Wide Web Consortium (W3C). The WAI have been active over many years in producing a structured set of guidelines representing 3 levels of compliance. These are referred to as level A, AA and AAA, with level A containing the minimum set of requirements and AA and AAA building deeper upon them. Each level is broken up into two parts: recommendations and requirements. It is not in fact possible to mandate and codify the measures necessary to be classed as 'accessible', as some of the issues cannot be assessed automatically and can only be assessed by human judgement - the selection of colour schemes, the judicious choice of alternative text for images and graphics, etc.
The NOF-digitise programme standards, for example, required all participants to achieve at least the WAI level A requirements. In practice it was found that most projects were able to fulfill all the demands and satisfy the guidelines to at least level A. Problems achieving compliance were mostly encountered with the use of JavaScript. We will therefore address some of these issue in some detail.
JavaScript is the predominant scripting language used to programme client browsers. JavaScript is a very useful tool, able to bring enhanced levels of functionality to pages and improve the visual interface in various ways - richer navigation schemes, selective layer visibility, mouse-triggered events and so forth. However, where these techniques become essential to the gain full access and use to the site, there is an accessibility problem. Level A compliance (and beyond) require a site to be fully accessible without JavaScript enabled. There are three reasons for making this demand:
Therefore, it is essential that pages are designed to fail gracefully when intended JavaScript functionality is not available on a client browser so that they do not result in the user being locked out from content or functionality on the site. It is necessary to distinguish between core and non-core uses of scripting: where scripting is used to enhance a page and does not result in any significant loss of experience in a non-JS enabled client, there is no real issue to resolve. However, where JavaScript has a core role on the page, for example in providing nested 'rollover' navigation devices, the non-JS client will not be able to fully access the site. This is not acceptable and other methods should be sought to provide the same level of functionality.
Due to the numerous applications to which JavaScript can be put to use on a page, it is difficult to produce effective advice covering all possible situations but the following general coding advice may be helpful in many common cases.
NOSCRIPT tags are very useful and should always be considered whenever a script block is inserted into a page. NOSCRIPT tags are ignored by browsers where scripting is enabled but interpreted when it is disabled. Combined with the use of 'document.write()' calls to insert HTML into the page, they allow areas of the page to be handled completely differently according to scripting capabilities. For example, if a polling script where to be inserted on a page that required JavaScript to function, the entire HTML necessary to display the polling utility on the page would be held within the SCRIPT tag and output using document.write(). The NOSCRIPT tag can then fill the same space with some alternate content.
The NOSCRIPT tag can also be useful to display a warning to a user, on the home page or elsewhere, notifying them that the site requires JavaScript to function properly or that some extra functionality may be lost if it is not enabled.
A common use of JavaScript is to control the display of large version images in pop-up windows. This is a good use of JavaScript but often causes problems. The following is advised:
EXAMPLE:
JavaScript dependent approach:
<a href="javascript:open_win('/img/large-image/y942_843_loa_85')">
<img src="/img/small-image/y942_843_loa_85.gif" alt=".." title="..">Click for larger photo </a>
Accessible approach:
<a onclick="open_win('/img/large-image/y942_843_loa_85'); return false;"
href="/img/large-image/y942_843_loa_85" alt=".." title="..">
<img src="/img/small-image/y942_843_loa_85.gif">Click for larger photo</a>
There are two main methods used to test a client's functionality in order to determine the (likely) functionality available and hence the compatibility of the browser with the JavaScript intended for the page.
Many scripts will parse the 'user agent' string written in to the browser to determine its type and version. However, this method may encounter problems as the user agent string value may be altered in any individual browser - some browsers give the user the ability to change or customise the user agent value, in other cases it is possible for a knowledgeable user to alter it - for example, the user agent value is held in a registry key in Windows Internet Explorer, and can be simply altered by editing the registry.
The preferred method to determine client-side functionality is to directly test for the desired capability. The required method / object / value is tested first and the code only executed if the test is passed. An (trivial) example:
if (document.images) {
//..script doing some manipulation on images in the document
}
else {
//..opportunity to inform user or quit script etc as appropriate
In this example, the block of code after the 'if' statement will only be executed if the test returns true, i.e. the Document Object Model (DOM) of the browser has an object called document.images. This style of coding will ensure that code will only be executed if the required functionality is available and is widely considered to be best practice, not only in client side JavaScript coding but in programming in general.
Where a site is unable to modify their design to deliver content in a way that does not rely on JavaScript, one solution is to offer a text-only version of the site.
The ideal way of achieving this is, conceptually, to maintain a separation between the content of the page and the presentation. This is not always easy to achieve and will usually have to be clearly specified at the time the system is commissioned, and will require a very well thought out system architecture and workflow process. Where this is attained, it should be straightforward to reformat the page content into multiple versions, using XSLT transforms, server-side templates or similar, therefore making it easy to provided an alternative, universally accessible version of a page alongside.
Page parsers can be also be used with some success. These utilities take the page code, strip out images and other extraneous items and produce a text-only version of the site. The results depend very much on the nature of the input page, and testing will be necessary to make sure that pages are satisfactorily rendered if this method is chosen.
Not recommended is the approach of maintaining a separate, text-only version of the site. Once content is stored in two separate, disconnected locations it can be a matter of time before the two become unsynchronised, usually resulting in the text-only version becoming out-of-date from the site it is intended to mirror.