Mass-unprefixing in Firefox 16
paulrouget.com
paulrouget.com
Dear browser makers, please, enable your prefixed CSS properties only in dev/night/beta builds. Leave only unprefixed CSS properties in the release stable browsers. With this you will be able to test new features, show them to web developers and, at the same time, save them from the madness of maintaining five different slightly different versions of new CSS properties. How many developers do you think will not remove the `-moz-` prefixes from they CSS? Who do you expect to say "from now on we will only support Firefox 16, let's drop the prefixed version used by older versions"?
Prefixes are for testing, not for stable applications.
[1] http://hsivonen.iki.fi/vendor-prefixes/ (see "Keep Experimental Features in Experimental Builds")
[2] https://groups.google.com/d/msg/mozilla.dev.platform/itl6mtx... (dbaron is Mozilla's representative to the W3C Advisory Committee as well being a principal engineer on the layout team and deeply involved in the development of CSS over the past decade.)
The CSS Working Group has agreed to give the go-ahead to browser implementers to unprefix CSS3 Transitions, Transforms, and Animations:
http://lists.w3.org/Archives/Public/www-style/2012Jun/0105.html
My point is that support for prefixed CSS properties should not be available in stable releases at all.
What they should have done from the beginning (and from other comments seems it will be done in the future) is to enable experimental prefixed CSS properties only in dev builds (to demonstrate to the W3C working group the feasibility of those new properties) and remove such properties from stable releases. In this way the stable releases will have _only_ W3C approved prefixes, not the current mess of draft + CR + TR support.
background: -webkit-linear-gradient(#fff, #fdd 75%, #fcc);
background: -moz-linear-gradient(#fff, #fdd 75%, #fcc);
background: -ms-linear-gradient(#fff, #fdd 75%, #fcc);
background: -o-linear-gradient(#fff, #fdd 75%, #fcc);
background: linear-gradient(#fff, #fdd 75%, #fcc);
Edit: And i should mention, this is not a mobile website, it's a desktop site.
Many people including myself think prefixes should be eliminated entirely[1], or replaced with a single prefix for the experimental[2].
[1] http://paulirish.com/2012/vendor-prefixes-are-not-developer-...
[2] http://www.quirksmode.org/blog/archives/2010/03/css_vendor_p...
background: -webkit-gradient(linear, 0 0, 0 100%, from(#fff) color-stop(75%, #fdd) to(#fcc));
And it's also an example of why prefixes were invented - the syntax is totally different.
I think vendors should unprefix the moment their prefix matches the spec.
most of the time when all prefixed version are the same it means there isn't a CR for the spec, and it is therefore in flux. unprefixing it would mean sites would break if and when it is changed before reaching recommendation.
It's kind of an obsession really, to the point that I furiously try not to use any third party library unless absolutely needed. And I haven't needed one until now. And I have also yet to regret these decisions.
I love JS though, in fact it's my favorite go to language for all kinds of work, heavy duty or light. I thinks that is also a part of the reason I am not inclined towards languages that compile to JS, they kind of destroy the experience of writing JS, which I am very fond of.
And I love JS too, but I use CoffeeScript since it's basically syntactic sugar for JS that saves me some keystrokes and makes my code more readable.
Oh well, i see, that makes sense, but as i said i don't use them so i don't know very much about them. I probably should use them once just for the sake of checking them out. And coffeescript too. Thanks!
As far as what these compilers do? Yes, CSS is nice enough as it is; that's why these compilers don't mess with the syntax too much. Most of what they're useful for is acting as a macro preprocessor (so you can use @color macros and automatically generate vendor prefixes, for instance), as well as for efficiently combining multiple source files.
If you're not familiar with how useful CSS compilers like LESS can be, I'd highly recommend that you take a look at Twitter Bootstrap. They use the macro-processing features pretty heavily, and to impressive effect. For instance, most of the colors in a Bootstrap site are configured in variables.less -- including light/dark color variants! -- so you can change around the whole color scheme of a Bootstrap site by adjusting just a few color definitions.
Yes, i stand corrected on that, i don't use them but that is not an excuse for ignorance. Plus compiling once makes more sense than compiling every time.
I also use webkit's web inspector for making live changes to a stylesheet and then saving the file in the end, very convenient compared to the changing-saving-uploading-refreshing cycle. Also it is another barrier to entry for my open source project. (someone has to know all the compiling process to contribute)
And as for uses of compilers, with CSS variables at the horizon (already functional in chrome canary IIRC) and other than this vendor prefixing problem (which everyone knows is very real and seeing the rapid release cycles of all major browsers now days, it is not far fetched to see a solution being cooked up some time soon) compilers don't have much to tackle. The cost v/s return ratio will reduce and probably even be negative in near future (if not already).
The cost is of compiling every time during development and once every time production code changes. I don't know about you, but if i am not live editing css in web inspector, i make a ton of very small changes to CSS and try them out one by one by refreshing the page. Not a great experience if i have to go to command line every time i change a byte.
> Efficiently combining multiple source files.
This problem can only arise in very big projects and if your project is in that category then there are all sorts of solutions available. (including writing your own compiler) But if that problem exists in a small or even middle sized project, them i believe, the solution is not the compiler but to re-factor the code base to use less files or merging them.
That's why, as was already mentioned, you generate them on the fly during development and generate them statically for production. That way you change the file on the file in your editor and it's updated immediately in your dev environment, but there's zero overhead in production.
Compass has a great command line feature called "watch" that simply watches for file system events (linux, os x) or polls (windows) your folders for changes and recompiles your stylesheets.
Vogue detects stylesheet changes right away and reloads them in your page via WebSockets.
Instead of dealing with Chrome's inspector, making changes, then copying them to your CSS (and translating if you like SASS), I simply make changes in my editor on one monitor while watching my site on the other monitor. The overhead is minimal - I see changes within about 500ms.
Compass is great even just as a build tool, and I used it before I learned SASS. It's fantastic for its simple mixins like linear-gradient and even better when you start chaining them together. One of my favorite mixins I use throughout my code:
@mixin buttonGradient($color){
@include basicBackgroundGradient($color);
&:hover, &.hover{
@include basicBackgroundGradient(hover-color($color));
}
&:active, &.active{
@include basicBackgroundGradient(active-color($color));
}
}
@mixin basicBackgroundGradient($color, $percent:12%){
background: $color;
@include background-image(linear-gradient($color, darken($color, $percent)));
}
If you go up, you can read my comment in which i say why i dont like extra libraries, compilers for web work. (http://news.ycombinator.com/item?id=4219980) This is pretty much why.
And for the global mixin, (Which is pretty cool IMO, it really enables a lot more possibilities and creativity) you pretty much give the reason not to use them anywhere out side of practice work. 18 lines are huge. Plus css variables are pretty much on horizon, it is well worth the wait.
CSS doesn't have variables. I can change my site's color scheme by changing just a few variables. (And I need to, too, because I'm not exactly great at picking them out in the first place...)
So i am not for reinventing the wheel, i just don't like to use square tires when round ones are better and readily available.
There is a proprietary filter property that works on previous versions IE.
that's why Mozilla is unprefixing IMO.
But yep, so much for standards.
I can live without gradients, my life will go on.