I believe your kbd example is incorrect. You suggest
> To save, press <kbd>Ctrl + S</kbd>.
But the spec (both W3C and WHATWG) suggests that individual keys should be nested inside an outer <kbd> tag: "When the kbd element is nested inside another kbd element, it represents an actual key or other single unit of input as appropriate for the input mechanism."
Thus, the example should be:
> To save, press <kbd><kbd>Ctrl</kbd> + <kbd>S</kbd></kbd>.
On the face of it, this seems ridiculous. It's too verbose, the tag name is misleading, and if you actually use the correct markup on GitHub or StackOverflow, it will render incorrectly because both sites assume the standalone <kbd> element represents physical keyboard buttons.
On the other hand, what's the value in semantic markup if we don't adhere to its semantics?
Practically speaking, I would be a happier person today if I hadn't read that part of the spec, and instead persisted on in blissful ignorance of the element's intended semantics. Thanks, specs.
Update: Looks like the <kbd> tag dates to the first draft of HTML 2.0 in 1993, where it was defined as "in an instruction manual, Text typed by a user." (https://tools.ietf.org/html/draft-ietf-iiir-html-00)
It remained more or less the same through HTML 4.01: "Indicates text to be entered by the user."
The notion of using nested kbd tags to indicate literal keys appears to have been introduced by WHATWG when working on HTML5.
...I'll dig into this more and actually write a blog post about it. I'm this close to relaunching a proper blog.
Talking about common usage of the kbd tag is kind of difficult given that it's never been commonly used, but personally, this was the only thing I ever saw it used for.
Edit: Oh, I missed that the example immediately following it is supposedly equivalent and only uses a single outer tag. I don't really understand then, if they're equivalent then what's the point of the defined semantics above?
2. <kbd><kbd>___</kbd></kbd> means "this specific, atomic keypress"
3. <kbd><kbd><samp>___</samp></kbd></kbd> means "this specific, atomic button/action/menu item"
So you can fall back to just using method 1, but it implies generic user input, inclusive of but not limited to keypresses. Thus, styling and treating a single kbd tag specifically like a keyboard button is still likely incorrect, since that precludes other semantically valid uses?
> Practically speaking, I would be a happier person today if I hadn't read that part of the spec, and instead persisted on in blissful ignorance of the element's intended semantics. Thanks, specs.
Hear hear! I found this hilarious and speaking to a shared feeling about some specs.
Of course, I kind of understand why these WTF elements make it into the specs. Maybe it's some stupid corner-case that the first implementor's code handled in an odd way because the code was simpler. Maybe there are other interactions that made it simpler to have this case be weird so that it would at least be consistent with that case. Most often, of course, it's because some vendor's code is a buggy mess of spaghetti code that can barely do the most basic stuff without falling over.
How is this updated (manually or automatically from official specs)?
Call me paranoid, but I see this diverging from actual specs, then people googling for "html reference" finding it and thinking it is something official. The result would be another W3Schools disaster[1].
In my opinion, the official W3 specification pages are not that bad, and alternatively there's the simpler MDN with strong community support (thus lower risk of deprecation).
Nice, but seems to be something wrong with the tagging. I unchecked all tags except `experimental` and I ended up with seven results, only one of which is actually experimental, (`picture`), the rest being _pretty_ cemented (`dt`, `li`, `option`, `td`, `th`, and `tr`).
It also seems to leave out some other expermental tags, like `wbr` and `slot`, that are on the site.
That's really well laid out. Less detail than MDN, but that site can be overkill.
Two small pieces of feedback. HTTPS support would be good. Also, when I scroll the list of element and then click on one, I'm taken back to the top of the list - I'd like to stay where I am.
I think a nice feature would be if the description pages linked to the relevant MDN page - for when you want more extensive information about an element.
Otherwise, I love this OP. Will likely use it regularly!
> I think a nice feature would be if the description pages linked to the relevant MDN page - for when you want more extensive information about an element.
This would be really good. Mostly I'm just looking for a quick reference, but sometimes I want all the details.
The checkbox filters at the top are somewhat un-intuitive. For example, If I uncheck everything but experimental, I want to see all experimental... but what I get is all experimental that are not block, inline, etc. I mean, as a developer, I see how those filters work... but as an end user, that isn't how I want them to work.
The <address> example is wrong. <address> defines contact information, but it's not appropriate for all postal and e-mail addresses.
http://html5doctor.com/the-address-element/
from a UX perspective i really like this. like the way it's laid out and presented - found it easy to scan.
I know a lot of people big up MDN vs W3Schools and all their arguments are basically correct but i find it (mdn) really ugly and hard to read. i often find myself going to w3schools to just copy a snippet or get a 1 line description of what i need.
mdn often feels more like a technical spec as opposed to a guide, which it kind of is i suppose.
While I do find this to be the case with regard to MDN it's worth noting that in many cases they have very good practical articles: Using XMLHttpRequest, Getting Started with WebGL, Using IndexedDB, etc.
This is really great. It would be even greater if it could be used offline. Instead of hitting the server every time I choose one of the elements to view.
Good stuff. Would be even nicer if it showed information about optionally self-closing tags, which is one of the main reasons I occasionally need to look at the spec. For instance:
> A p element's end tag may be omitted if the p element is immediately followed by an address, article, aside, blockquote, div, dl, fieldset, footer, form, h1, h2, h3, h4, h5, h6, header, hgroup, hr, main, nav, ol, p, pre, section, table, or ul, element, or if there is no more content in the parent element and the parent element is not an a element.
Without compatibility information this is more dangerous than useful. For example, a naive HTML author would use the semantic tags, only to fail in most browsers. There is mention neither of compatibility nor polyfill. Better off with MDN and caniuse.com.
Nice...would be useful to have the summary of what a tag's purpose is on the main page list of tags. For three letter or less tags it's not always obvious what I'm looking to achieve a specific task.
In general this is pretty good. One mistake that it doesn't make is to use low contrast for everything.
However, the code examples don't look very nice on my box due to the font which comes out as Nimbus Mono L. How about you link to an actual font? (I like Ubuntu Mono, though there are lots of other good programming fonts).
Also with respect to code samples, would black text on a white background be a possibuility?
On the font issue: you're effectively saying "I don't like the default monospace font my browser is configured to use, please change your website to fix this" (since Nimbus Mono isn't in their CSS font family list). I feel you as a user have some responsibility in that case.
> you're effectively saying "I don't like the default monospace font my browser is configured to use, please change your website to fix this"
No, that's not true. I have just changed the monospace font on Firefox to Ubuntu Mono, and it is still showing as Nimbus Mono. This is the website doing this -- for example the change is reflected fine in HN.
> I feel you as a user have some responsibility in that case.
These days when I build websites I normally specify specific web fonts so that I know it won't default to something not good.
So either it is using one of those fonts (Courier and Nimbus Mono L are similar I guess), the site is serving you special CSS that I'm not getting (and I've tested it on Linux and Windows), or Firefox has done some weird font aliasing and one of the above is being treated as Nimbus Mono L.
I love things like this, but I find I have about 30 of these sorts of things bookmarked, and I never think to pull them up when I'm in need – I immediately go to Google and find something like the Mozilla ref @shpx mentioned. It's a real problem I wish there was a solution for. Other than of course having a better memory.
The filter checkboxes are a bit weird, for example elements (such as br and img) which are both self-closing and inline only show up if both self-closing and inline are selected, instead of one or the other. Also there are several elements such as li which are not categorized and show up even if no checkboxes are selected.
To expand, the issue seems to happen after clicking a tag and then pressing back. It's not specific to meta. This time block elements were appearing when not tagged.
And as an example, when filtering by "self-closing" the <img> tag does not appear. That happens because "inline" is not selected. Same behavior with <embed> and removing either "self-closing" or "block".
quite cool! but even after deselecting all the options, I'm still being shown some tags (select, tr, th, td, dt, li). I think it's because they haven't been tagged with anything
Every time I look at lists of HTML5 elements, I about how I really need to get better at writing semantic HTML. Tomorrow I will go to work and create 500 more div tags.
If I'm supposed to let them sell my privacy and attention, I wouldn't really call it free. Money isn't the only thing of value I can give to a website.
It's the voluntary nature of the payment and whether the cost is tangible to you that determine what is free, not the currency used. Substitute "your FBI file" for your ad network profile and "compute time on your Amazon instance" for JavaScript time in your browser to make this clearer.
Same thing but guaranteed to be up to date and (more) complete.
For example, htmlreference.io's page for <input> doesn't mention the autocomplete attribute. MDN lists all its possible values.
http://htmlreference.io/element/input/
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/in...