Change HTML Anchors to IDs (Named Anchors Aren't Valid HTML5)

[tajmone]

Currently, the --anchors options attaches a name attribute to <a> tags in linenumbers:

<a name="l_4"></a><span class="hl lin">  4 </span>  <span class="hl kwb">INC</span>

In HTML5 named anchors are not valid tags:

Differences Between HTML 4.01 and HTML5

In HTML 4.01, the tag could be either a hyperlink or an anchor. In HTML5, the tag is always a hyperlink, but if it has no href attribute, it is only a placeholder for a hyperlink.

HTML5 has some new attributes, and some HTML 4.01 attributes are no longer supported.

--- https://www.w3schools.com/tags/tag_a.asp

The <a> name attribute is not supported in HTML5. Use the id attribute instead.

--- https://www.w3schools.com/tags/att_a_name.asp

Highlight should drop the <a name=""> tag in favour of placing an Identifier directly on the line-number span (hl lin):

<span class="hl lin" id="l_4">  4 </span>  <span class="hl kwb">INC</span>

... or at least it should offer this alternative for HTML5-specific output --- it seems to me that using a <span id=""> would be valid also in HTML4:

-- https://www.quackit.com/html/html_4/tags/html_span_tag.cfm

But if for some backward compatiblity reasons (or to preserve output consistency in existing projects) the current line-numbers anchoring system should be preserved, it might be worth considering an additional option to target HTML5 specifically --- the introduction of this switch might also be a chance to introduce a more HTML5 compliant way of wrapping code blocks with <pre><code> tags (plus attributes targetting computer languages).

But for the <pre><code> tags problem, the --fragment options offers a workround. Not so for the line-numbers anchoring system, on which the users doesn't have much control (except, maybe, via some custom plugin).

HTML4 is becoming obsolete, and I think that Highlight should target valid HTML5 output via it's default options, so that its usage within projects should not affect the validity of their HTML5 syntax.

With the recent update to v2.0, pandoc changed the way it syntax highlights code in HTML (line numbers were previously achieved via tables, now dropped in favour of divs and spans with Ids), to catch up with HTML5 standards. Maybe pandoc could be a good reference point on how (X)HTML code should be formatted, as much thought and discussion as been going on during the last year regarding which HTML5 tags and attributes should be used for syntax highlighting code blocks. Of course, this forced pandoc to bump it's major version to 2.0, effectively marking a point bracking backward compatibilty --- but not doing so was causing more trouble to HTML5 complaint projects than updating existing scripts and templates.

For Highlight the case shouldn't be one of breaking backward compatiblity, if a new HTML5 format is added on top of the current HTML(4). But introducing this new format might require some careful thoughts, so that it's implemented at once with all it's requirements, or either gradually (as an experimental feature) and gets fine-tuned through actual usage and feedback.

Of course, changing the way line-numbering is achieved might have repercussions on CSS themes (but these could also comply to a new HTML5 option accordingly). The same could be said for standalone HTML5 documents, which could benefit from a newer HTML5 complaiant template.

Needless to say, I'm willing to contribute on this front.

[andre-simon]

There should be no issues if <a name='xxx'> is replaced by an ID within a span, as referencing and using these links do not change. If you apply -O xhtml the attribute name is id instead of name. I forgot the name when I chnaged the default HTML output towards HTML5.

[andre-simon]

Done in v3.41.