- From: Nils Dagsson Moskopp <nils@dieweltistgarnichtso.net>
- Date: Sat, 16 May 2015 22:40:16 +0200
- To: Taylor Hunt <taylorcharleshunt@gmail.com>, whatwg@whatwg.org
Taylor Hunt <taylorcharleshunt@gmail.com> writes: > Hello, > > I was wondering about the best way of indicating a <code> element's > contained computer language for syntax highlighting. The > specifications are quite clear that using @lang for that is abuse, so > many existing implementations use a data-lang attribute, class name, > or something similar. > > I was curious if repurposing the "type" attribute onto <code> elements > for this purpose would be a good idea. The <script> and <style> > elements already use it to indicate what language they contain, so > there would seem to be precedent. (An argument against this would be > <input> already overloading type's meaning, but that ship may have > sailed.) Including non-standard languages within these tags with a > differentiating value for type is widely-practiced, such as for HTML > templating, or more exotic implementations like in-browser > interpreters or thegrid.io's use of <style type="text/gss">. > > Using code[type] would have the advantage of an existing vocabulary > for unambiguously indicating a code language, through MIME types. It > also works today, and should have no problems in older browsers. This proposal would also have the benefit of making it possible to automate generation of the type attribute contents using file(1) or magic(5), which makes it easy to annotate code elements automagically. > The most obvious benefit would be syntax highlighting, but there could > be other use-cases: more intelligent enunciation by speech > synthesizers, for example. It would also make it easier to search for source code examples in a specific language among a large corpus of content. I would love this! -- Nils Dagsson Moskopp // erlehmann <http://dieweltistgarnichtso.net>
Received on Saturday, 16 May 2015 20:41:01 UTC