A big problem I experience is the lack of web-safe keyboard shortcuts for complex web-apps. If looking for a safe key-combination, you have to consider the superset of OS and Browser combinations and that leaves almost nothing.

The gmail approach is for unmodified keys but that only works if you don't have focus in an editable field. The gdocs approach is for menu chords but that sucks for frequent actions and still generates weird dual handling of events by the browser.

I'd love a web/browser standards committee to preserve the universal utility of the keyboard. It could ring-fence some application specific shortcuts, give a semantic abstraction of some actions e.g. undo, and standardise mechanisms for os specific remappings.

As it is, as more tech comes along, the less usable keyboards become. I am still dumbfounded that the iPad lacks a tab button or any chord for one despite claiming to be usable for word-processing. This basically means no web-app can ever assume a keyboard has a tab key. Gah.

This is especially painful since sometimes browsers have different built-in shortcuts, and sometimes the in-browser shortcuts differ cross-platform.

Also, coming from Windows/Linux world, I was surprised how many physical keys are missing on a Mac keyboard.

Getting an error on that page in Chrome.

    Refused to execute script from 'https://rawgithub.com/dmauro/Keypress/master/keypress.js' because its MIME type ('text/html') is not executable, and strict MIME type checking is enabled.

EDIT: It's because Keypress is using rawgit.com instead of cdn.rawgit.com to host the actual keypress.js script.

Details here - http://rawgit.com/dmauro/Keypress/master/keypress.js

Working script source - http://cdn.rawgit.com/dmauro/Keypress/master/keypress.js

RawGit is refusing to send the page because you're not supposed to use it for production sites without using the cdn.rawgit.com link.

To be more precise: by looking at the response in Firebug's Net panel I see

<h1>Blacklisted for abuse</h1>

<p> This request has been blacklisted for sending too much traffic to <code>rawgit.com</code>. Please contact the owner of the site that sent you here and ask them to use <code>cdn.rawgit.com</code> instead, which has no traffic limit. </p>

Nevertheless I wonder if this will work when Firefox's search as you type is enabled. I guess the browser should eat up all those events, but maybe I'm wrong. Can't check it because of the blacklisted script and no time to write my own test now.

Seems like they also deliver this with an HTTP 200 status. So Firefox actually tries to parse the error page as JavaScript.

Previous discussion https://news.ycombinator.com/item?id=6464138

    Keypress: A Javascript library for capturing input (dmauro.github.io)
    192 points by nsmalch 468 days ago | flag | cached | share | comments

I thought only one submission per url was possible

Per year, I think.

From the FAQ [1]:

"Are reposts ok?

If a story has had significant attention in the last year or so, we kill reposts as duplicates. If not, a small number of reposts is ok."

As with much of the HN system, that's intentionally vague, but in my interpretation, reposts can occur more frequently than once per year.

[1] https://news.ycombinator.com/newsfaq.html

An alternative is mouestrap: http://craig.is/killing/mice

I've used it successfully in Marionette and Angular applications.

It is 29k unminified, vs 35k of keypress.js.

Plus one for Mousetrap. I have used it too...

Also, it has a better name. ;)

I'd like to see scoping added so it's possible to easily switch keyboard bindings based on the feature/context.

https://github.com/madrobby/keymaster does an OK job with this, and too bad it's not being actively maintained.

The first param of the constructor for the Listener is an element to be bound to so it will only listen for key events for that element.

More aimed at game developers, but for anyone who wants a super lightweight subset of this functionality.

https://github.com/danprince/keystate

Now with Konami code!