Support for HTML5 boolean attributes

[dalgard]

I haven't gone into this extensively, but you should at least consider adding autofocus to your list of boolean attributes. Thanks :)

[kangax]

autofocus, multiple, required, hidden, contenteditable, spellcheck. Any other ones? Don't see anything else

/cc @mathiasbynens

[dalgard]

Looks reasonable - awesome.

[mathiasbynens]

For future reference, the current list of boolean attributes in this HTML minifier is the following:

"checked,compact,declare,defer,disabled,ismap,multiple,nohref,noresize,noshade,nowrap,readonly,selected"

autofocus, multiple, required, hidden, contenteditable, spellcheck. Any other ones?

contenteditable and spellcheck aren’t proper boolean attributes, although they do get a default value other than the empty string.

Hmm, async, itemscope, and seamless come to mind, but there are probably more…

[mathiasbynens]

Update:

$ curl -s "http://www.whatwg.org/specs/web-apps/current-work/" | grep -C 3 "boolean\s*attribute"

reversed, allowfullscreen, typemustmatch, muted, sortable, novalidate, formnovalidate, open, default, inert, truespeed (along with all the others mentioned in this thread). There may be even more.

Here’s the full list so far:

allowfullscreen
async
autofocus
checked
compact
declare
default
defer
disabled
formnovalidate
hidden
inert
ismap
itemscope
multiple
multiple
muted
nohref
noresize
noshade
novalidate
nowrap
open
readonly
required
reversed
seamless
selected
sortable
truespeed
typemustmatch

For any of these attributes, attributename="any-value" can be optimized into just attributename.

Then, there are a few non-boolean attributes that take a default value instead of the non-empty string:

contenteditable
spellcheck

Variations of contenteditable="true" and contenteditable="" can be optimized into just contenteditable. Any other values for this attribute that are not equal to the literal string false map to the default state, meaning e.g. contenteditable="lolwat" or contenteditable=" true" (with leading space) can be optimized by just removing the attribute completely.

Variations of spellcheck="true" and spellcheck="" can be optimized into just spellcheck. Any other values for this attribute that are not equal to the literal string false map to the default state, meaning e.g. spellcheck="lolwat" or spellcheck=" true" (with leading space) can be optimized by just removing the attribute completely.

[kangax]

Thanks @mathiasbynens. So we're 19 attributes short now.

[ ] allowfullscreen
[ ] async
[ ] autofocus
[x] checked
[x] compact
[ ] contenteditable
[x] declare
[ ] default
[x] defer
[x] disabled
[ ] formnovalidate
[ ] hidden
[ ] inert
[x] ismap
[ ] itemscope
[x] multiple
[ ] muted
[x] nohref
[x] noresize
[x] noshade
[ ] novalidate
[x] nowrap
[ ] open
[x] readonly
[ ] required
[ ] reversed
[ ] seamless
[x] selected
[ ] sortable
[ ] spellcheck
[ ] truespeed
[ ] typemustmatch

[mathiasbynens]

contenteditable is not a boolean attribute. Updated the list.

[kangax]

It could be an empty string (or empty attribute), according to this

[mathiasbynens]

True, that’s what it says in the WHATWG HTML spec too (“The empty string and the true keyword map to the true state”). spellcheck isn’t a boolean attribute either, although it has similar behavior.

Just saying, we’re just not looking for boolean attributes here, but also for enumerated attributes that get a default value instead of the empty string. My grep trick definitely didn’t catch all of these.

[evilpie]

This might also be interesting. Most of the boolean attributes in Firefox. http://mxr.mozilla.org/mozilla-central/search?string=attribute+boolean&regexp=1&find=%2Fdom%2Fwebidl&findi=&filter=^[^\0]*%24&hitlimit=&tree=mozilla-central (Still misses DOMWindow or something)

[mathiasbynens]

@alrra found another attribute that can be optimized: draggable. It’s not a boolean attribute, but its missing default value is auto. So, the HTML minifier could optimize e.g. draggable="auto" into just draggable.

[jefffis]

How's about the scoped attribute for the style tag?

[tobaco]

download ?

[jefffis]

I think download needs a string pointing to the actual downloadable file.

[ahmedelgabri]

autocorrect & autocapitalize?

[evilpie]

For download the value is an optional filename. So just download still forces a download of the resource specified by href. (I implemented this feature in Firefox)