include-pattern: Difference between revisions
PaulWilkins (talk | contribs) m (Consistant naming of author in examples) |
(fixed quick ref to use 'title' rather than content, removed unnecessary misconception about misconceptions#current_behaviors_and_usage_patterns_in_general, added sub-sub hdngs for linklist vs tabs) |
||
Line 18: | Line 18: | ||
The include-pattern is a mechanism to include content from one microformat into another microformat elsewhere in the same document, using hyperlinks (recommended) or OBJECT. For example: | The include-pattern is a mechanism to include content from one microformat into another microformat elsewhere in the same document, using hyperlinks (recommended) or OBJECT. For example: | ||
<pre><nowiki><a class="include" href="#author" | <pre><nowiki><a class="include" href="#author" title="James Levine"></a></nowiki></pre> | ||
<pre><nowiki><object class="include" data="#author"></object></nowiki></pre> | <pre><nowiki><object class="include" data="#author"></object></nowiki></pre> | ||
Line 84: | Line 84: | ||
===== accessibility ===== | ===== accessibility ===== | ||
There are two points of view on this. <span id="Accessibility_concerns">The first is that the user experience of Assistive Technology can(might?) be severely degraded by the presence of hyperlinks which do not contain text ('''citation needed''').</span> The second is that the first claim is only a side effect of some Assistive Technologies extracting the hyperlinks in a document and presenting them as a list, and that such technologies / features have no right to do so, and therefore it is not your problem to worry about. See [http://joeclark.org/appearances/atmedia2007/#context Joe Clark: When accessibility is not your problem: Headings and links read out of context], in particular: | <strong>Hyperlinks presented as an extracted list.</strong> There are two points of view on this. <span id="Accessibility_concerns">The first is that the user experience of Assistive Technology can(might?) be severely degraded by the presence of hyperlinks which do not contain text ('''citation needed''').</span> The second is that the first claim is only a side effect of some Assistive Technologies extracting the hyperlinks in a document and presenting them as a list, and that such technologies / features have no right to do so, and therefore it is not your problem to worry about. See [http://joeclark.org/appearances/atmedia2007/#context Joe Clark: When accessibility is not your problem: Headings and links read out of context], in particular: | ||
<blockquote><p>As writers, we are not authorizing you or Jaws to pull out our link text and remix it. Why don’t you rearrange the sentences, too?</p></blockquote> | <blockquote><p>As writers, we are not authorizing you or Jaws to pull out our link text and remix it. Why don’t you rearrange the sentences, too?</p></blockquote> | ||
<strong>Hyperlinks and tab focusing in common browsers</strong>. As noted in the [http://microformats.org/wiki/include-pattern-feedback include-pattern feedback page], any visible hyperlink used with an <code>href</code> attribute ends up in the browser's tab cycle. This affects all keyboard users (both sighted users that can't/don't operate a mouse, or screen reader users in the extreme case). Even empty hyperlinks receive focus. This can be tested by modifying the code of the original [http://allinthehead.com/demo/include.html a@include test page] to visually follow the tabbing from one hyperlink to the next (using CSS or, for test purposes, subtly changing the <code>href</code>, tabbing through the page, and observing the change in the URL displayed in the browser's status bar). Hyperlinks hidden with CSS <code>display:none</code> do not end up in the browser's tab cycle. | |||
Thus it is recommended that Hyperlink Include Pattern hyperlinks lacking inline content {{should}} be hidden with CSS. As they have no content, this stylistic change has no net effect on the document's content semantics from a user's perspective. | |||
=== Object === | === Object === |
Revision as of 23:18, 6 November 2007
Include Pattern
Initially developed as part of resume-brainstorming, the include pattern is a mechanism to include a portion of data from one area of a page into another area of the same page. The following is documentation for re-use of the pattern in other microformats, and for publishers working with it.
- Editors
- Tantek Çelik
- Ben Ward
background
hResume needed the ability to include a name from one hCard at the top of a resume — the person's contact details — into into the separate hCards used in the same person's employment history. Repeating so much data would be inconvenient to publishers, irritating to consumers and would not have matched the existing publishing techniques used in Resumes and Curriculum Vitæ. The include pattern is a mechanism to reference data from the same page, avoiding repetition.
scope
The include pattern is strictly limited to the scope of the current page. It cannot be used to include content from other URLs.
quick reference
The include-pattern is a mechanism to include content from one microformat into another microformat elsewhere in the same document, using hyperlinks (recommended) or OBJECT. For example:
<a class="include" href="#author" title="James Levine"></a>
<object class="include" data="#author"></object>
In specs which cite the include-pattern, either of the above code snippets will cause a microformats parser to replace the A or OBJECT element with class name "include" with the content fragment with ID "author". Full examples follow.
in general
To reference includes, use an include element with class name "include" and a document fragment identifier. The content to be included should have an ID attribute set, and that ID should be referenced from the HREF or DATA attribute at the point of inclusion.
The include element with class name "include" indicates a reference to a sub-tree elsewhere in the document which must be included in-place by microformat parsers. That is, the element with class "include" is _replaced_ in the DOM by the referenced sub-tree.
To prevent infinite loops, if a class="include" refers to itself or to an ancestor in the parse tree, then it is ignored and has no effect on the parser.
Per the scope, the OBJECT 'data' attribute and hyperlink 'href' attribute MUST be local ID references when used as include pattern instances. External references (requiring a consuming application to load an external resource) are not supported by this method.
There are two HTML elements available to reference includes, hyperlink/Anchor and OBJECT. They are documented below.
These methods of property indirection via a hyperlink element can apply to any/all properties in class-based microformats, but should only be used where a microformat explicitly states that the include-pattern is a dependency. For example, XOXO does not reference the include-pattern at this time, so sub-trees cannot be included by reference in XOXO. hResume and hReview do reference the include pattern.
Hyperlink
The recommended way to reference includes within microformats is to use a hyperlink.
Hyperlink example
Here is an hCard from the beginning of a resume, shown here as a verbose hCard.
<span class="vcard"> <span class="fn n" id="jl-name"> <span class="given-name">James</span> <span class="family-name">Levine</span> </span> </span>
Elsewhere on the page, a second hCard re-uses the "fn n" content from the first hCard:
<span class="vcard"> <a href="#jl-name" class="include" title="James Levine"></a> <span class="org">SimplyHired</span> <span class="title">Microformat Brainstormer</span> </span>
A microformat parser must treat the second hCard as follows, with the hyperlink include element completely replaced (including inner-text) by the sub-tree that was referenced:
<span class="vcard"> <span class="fn n" id="jl-name"> <span class="given-name">James</span> <span class="family-name">Levine</span> </span> <span class="org">SimplyHired</span> <span class="title">Microformat Brainstormer</span> </span>
notes and issues
Note: The id
attribute value in the example "jl-name" was chosen for demonstration purposes. The text of the value has no semantic per the include-pattern other than to connect the source of the include to the destination. Authors SHOULD use good POSH techniques to choose id
and class
names.
Using the hyperlink include pattern causes no known issues in any web browser, apart from keyboard access problems noted below.
Authors SHOULD supply content text or at least title
attribute text for the hyperlink itself. This can require repeating a small piece of information (such as a person's name in an hCard), or including generic text suitable for the context of the page.
accessibility
Hyperlinks presented as an extracted list. There are two points of view on this. The first is that the user experience of Assistive Technology can(might?) be severely degraded by the presence of hyperlinks which do not contain text (citation needed). The second is that the first claim is only a side effect of some Assistive Technologies extracting the hyperlinks in a document and presenting them as a list, and that such technologies / features have no right to do so, and therefore it is not your problem to worry about. See Joe Clark: When accessibility is not your problem: Headings and links read out of context, in particular:
As writers, we are not authorizing you or Jaws to pull out our link text and remix it. Why don’t you rearrange the sentences, too?
Hyperlinks and tab focusing in common browsers. As noted in the include-pattern feedback page, any visible hyperlink used with an href
attribute ends up in the browser's tab cycle. This affects all keyboard users (both sighted users that can't/don't operate a mouse, or screen reader users in the extreme case). Even empty hyperlinks receive focus. This can be tested by modifying the code of the original a@include test page to visually follow the tabbing from one hyperlink to the next (using CSS or, for test purposes, subtly changing the href
, tabbing through the page, and observing the change in the URL displayed in the browser's status bar). Hyperlinks hidden with CSS display:none
do not end up in the browser's tab cycle.
Thus it is recommended that Hyperlink Include Pattern hyperlinks lacking inline content SHOULD be hidden with CSS. As they have no content, this stylistic change has no net effect on the document's content semantics from a user's perspective.
Object
The Object Include Pattern is semantically superior to the Hyperlink Include; it is being used to embed content into the page. The object
element based include was the original developed include pattern. However, there are serious browser compatibility issues that can affect some implementation scenarios and thus the above Hyperlink Include Pattern was developed and is now preferred.
Object example
Here is the same hCard from the beginning of the resume in the previous example.
<span class="vcard"> <span class="fn n" id="jl-name"> <span class="given-name">James</span> <span class="family-name">Levine</span> </span> </span>
Elsewhere on the page, a second hCard re-uses the "fn n" content from the first hCard:
<span class="vcard"> <object data="#jl-name" class="include"></object> <span class="org">SimplyHired</span> <span class="title">Microformat Brainstormer</span> </span>
A microformat parser must treat the second hCard as follows, with the object
include element completely replaced by the sub-tree that it referenced:
<span class="vcard"> <span class="fn n" id="jl-name"> <span class="given-name">James</span> <span class="family-name">Levine</span> </span> <span class="org">SimplyHired</span> <span class="title">Microformat Brainstormer</span> </span>
This method of hCard property indirection via an object element can apply to any/all properties in class-based microformats.
notes and issues
- Unlike the hyperlink pattern, the Object Include Pattern does not require any textual fallback. This makes it invisible to Assistive Technology with no adverse affect.
- Apple's Safari (WebKit-based) browser has some rendering issues with this use of
object
: You should set a width and height of "0" for the include elements to resolve this (version(s) affected needed). - Bugs have been reported in some web browsers (Internet Explorer, Safari, Firefox) that
object
elements referencing fragments of the same document erroneously cause the browser to make additional HTTP requests (version(s) affected needed). For scenarios where HTTP requests are a sensitive performance measure, this makes the Object Include Pattern inappropriate. - Versions of Microsoft Internet Explorer with ActiveX controls disabled will throw a warning prompt upon finding an
object
element (version(s) affected needed).
acknowledgements
Thanks to discussions and brainstorms with a bunch of folks: Ryan King, James Levine, the whole crowd at the Microformats specifications working session at MashupCamp, Brian Suda, Scott Reynen, Drew McLellan.
specifications using
considering
All class-based microformats SHOULD consider using and explicitly normatively stating support for the include pattern.
implementations
- X2V implements the include-pattern when parsing hCards and hCalendars for both object.include and a.include.
references
normative
- HTML 4.01
informative
- the include pattern
- include-pattern examples in the wild - an on-going list of websites which use the include pattern.
- include-pattern FAQ - if you have any questions about the include pattern, check here, and if you don't find answers, add your questions!
- include-pattern feedback- general feedback (as opposed to specific issues).
- include-pattern issues - specific issues with the include pattern.
- include-pattern strawman - alternative proposals
- See also blogs discussing this page.