Stubbornella - Nicole Sullivan

Last week, Stoyan Stefanov and I spoke at Velocity Conference about optimizing massive CSS. We talked about our experiences optimizing large-scale sites like Facebook and Yahoo!, and we discussed our findings regarding the CSS efficiency of the Alexa Top 1000 websites.

Velocity was kind enough to share videos of the session.

What is the state of the internet regarding CSS performance? Kind of sad. We aren’t getting a lot of the basics right, and when we look at the more advanced techniques, there are some spectacular examples of what-not-to-do. Why do we care about CSS performance? As Stoyan talks about in the beginning of the video, it blocks progressive rendering and it is very difficult to auto-minify.

What is the internet made of? At least the UI layer is mainly composed of media blocks. I talked about the Facebook stream story before, and all the tiny objects of which it is composed. For the most part, the stream story is made up of the media object repeated over and over.

The media object is an image to the left, with descriptive content to the right, like this Facebook story:

It is a very simple object, but it is very powerful. We can eliminate many lines of code abstracting this repeating pattern. The code for the media block and many other “web Lego” are available on the Object Oriented CSS open source project.

In my previous post, I said:

Shoehorning CSS and HTML into PHP abstractions prevents the code from being DRY and ultimately leads to code bloat, because, the CSS and HTML require a far more granular object structure than the PHP.

And then I didn’t expand on it, or give proper context. Ooops, sorry!

In the PHP layer, the following stream story (from facebook) might be a single object. All of the logic of determining which bits of HTML to show or hide are contained within one object structure.

Frequently, developers try to make the CSS match this middle-end logic. We expect all the CSS code for the stream story to be sand-boxed within a wrapper id, for example #story. This ends up with bloated code that isn’t DRY. Why? because the proper architecture for the CSS layer is much more granular. Trying to base the CSS architecture on the PHP layer is something akin to trying to use the DB schema to configure apache. It just doesn’t work.

Each layer in the web stack has its own architecture. You wouldn’t expect the DB schema to be used to architect the PHP middleware, and yet people expect that of the HTML and CSS. HTML needs to be written with a sense of the meaning of the data or content, something I call code semantics. Code semantics are incredibly important in the HTML for both portability and accessibility. On the other hand, CSS really is a separate layer in the stack, and it needs its own architecture that reflects the visual semantics of the page.

Visual semantics describe all the repeating patterns in the design of the page. They represent the fundamental building blocks of your website. In fact, they are often portable across sites with only minor modifications. Visual semantics shouldn’t necessarily be tied to HTML semantics, because you want an architecture that fits the unique requirements of each layer of the stack…

Don’t get me wrong, I think CSS is awesome. It is a great way of defining the UI, but it could be even better. I’m excited about the special effects, transitions, and graphic elements currently being added to the CSS specification. They will help us write faster pages by eliminating the need for UI graphics for things like rounded corners. On the other hand, we also need to add structure to the language to make it easier for designers and developers to author new pages and applications. I hope the CSS Working Group will consider my suggestions.

Suggestions for the CSS Working Group

The community has been talking about some form of constants or variables in different incarnations for almost a decade, it is time to make this a reality. While it is possible to duplicate this functionality with a preprocessor (an excellent stop gap until browser support catches up), ultimately this is a tool which should live within the CSS itself. It is a powerful way of expressing more about the objects we are building.

The mixins and prototypes (and associated includes and extends properties) are designed to allow document authors to abstract reusable bits of code and better manage and maintain their style sheets. The goal is to mimic the effect of using multiple class names in the HTML without the drawbacks associated with current implementations. Authors need extends and include in order to write faster, smaller, more efficient style sheets.

CSS is a powerful expressive language. It needs certain modifications so that it will be robust, maintainable, and easy to implement.

From JSConf.eu

I recently went to New York to hang with some people who are interested in HTML5 and figure out what I thought about the future of this web standard. I’m a skeptic by nature, so I went into our little quest expecting to be unimpressed by HTML5, but in fact, it isn’t so bad, and even has a few additions I’m excited about.

Down with Pseudocode!

On the other hand the spec itself drives me crazy because I feel like pseudocode is a poor substitute for properly and clearly stating what you are trying to achieve. It is easy to mask your agenda in pseudocode and harder for people to sort out later what was intentional versus incidental.

#html5 .pseudo-code { display: none;} Thought experiments don’t belong in a spec & pseudocode can’t replace properly specifying requirements.

@stubbornella

Speaking the CSS Summit today, I listened as people voiced concerns about browser support and bugs. It might surprise my fellow CSS developers, but many of the seasoned, bearded (ok, I’m making this part up) engineers working on the innards of browsers and rendering engines may not actually write much CSS themselves. They need our help.

We need to tell them what is broken, via good bug reports with steps to reproduce, version information, and relevant comments, screenshots and ideally test cases. Can you boil the bug down to the simplest possible version of itself? How little code do you need to trigger the rendering problem? That’s the test case you want to submit.

A marked change has occurred since the first Velocity Conference a year ago, and while the effects are not yet obvious, they will be. The web is still slow, but we have something now, that we didn’t a year ago: business metrics. This was the year we quantified the impact of performance choices on our businesses, and the results were astounding.

For those of you who couldn’t attend, this article summarizes results from AOL, Shopzilla, Google, Bing, Hotmail, and Chrome and gives pointers to highlights of Velocity Conference 2009.

Today is a great day for Smush it because Yahoo! decided to accept it into their family of tools. This will allow the tool to run on fanatically maintained servers, with the Yahoo! style quality of service. It doesn’t get any better than that. I feel like my baby is all grown up and going off to college!

Want to Optimize Images like a Pro?

Stoyan and I have contributed a chapter to Even Faster Web Sites, that outlines all the fantastic tricks you can use to make your sites smaller. We hope you find it useful, and of course we welcome feedback! It will show you how to make image optimization part of your own system, and give you all the information you need to make images as small as possible.

I’ve been tweeting and posting to delicious about reflows and repaints, and it seemed time for a blog post. Opera lists repaint and reflow as one of the three main contributors to sluggish JavaScript, so it definitely seems worth a look.

Reflows are very expensive in terms of performance, and is one of the main causes of slow DOM scripts, especially on devices with low processing power, such as phones. In many cases, they are equivalent to laying out the entire page again.

How to avoid reflows or at least minimize their impact on performance?

1. Change classes on the element you wish to style (as low in the dom tree as possible)
2. Avoid setting multiple inline styles
3. Apply animations to elements that are position fixed or absolute
4. Trade smoothness for speed
5. Avoid tables for layout
6. Avoid JavaScript expressions in the CSS (IE only)

Tools

A few tools have made waves lately. Stoyan Stefanov and I have been looking for decent ways to measure reflows and repaints and there are a few tools which show promise (despite being very early alpha).

• UPDATE: Lindsey Simon at Google wrote a bookmarklet that tests reflow times on any browser. Fantastic. (Note: all that shaking is normal!)
• John Resig wrote a bookmarklet to visualize paint events.
• Kyle Scholz created this tool to visualize paint events before onload.
• Alex at Yoono has created the XUL profiler.

Has anyone else seen any cool tools for evaluating reflows?

Yahoo! Developer Network has released a video of my Object Oriented CSS talk at Web Directions North just in time for Ada Lovelace day. I’ve also been included in a feature on Women in Technology. I’m absolutely flattered to be included among these fantastic technical women. Wow.

Object Oriented CSS: for high performance websites and web applications.

Find out more about object oriented css

1. Open source project on github (GIT is having some DNS issues, be patient)
2. Follow along with the slides on slideshare
3. Join the OOCSS google group

My Object Oriented CSS grids and templates are open sourced on github. They have all the functionality of YUI grids plus some important features.

• Only 4kb, half the size of YUI grids. (I was totally happy when I checked the final size!)
• They allow infinite nesting and stacking.
• The only change required to use any of the objects is to place it in the HTML, there are no changes to other places in the DOM and no location dependent styling. Eases back-end development and makes it a lot easier to manage for newbies.
• Solution for sub-pixel rounding errors.

http://wiki.github.com/stubbornella/oocss

template.css and grids.css

…My prediction is that you’ll be writing complex layouts in less than 24 hours without adding a line to the CSS file.

A bit of a rant

The cascade is something like a new data structure, and the ways for dealing with it are algorithms you never learned in school. You couldn’t have, because traditional engineering school poo-poos the front-end and web engineering in favor of stale (but still valuable) traditional software engineering.

Perhaps you realized this web thing wasn’t going away, so you learned a bit of javascript. That went OK, how hard is it to add another language to your repertoire? Now you just wish it would compile like a proper language. You hate the browsers, and (foolishly) wish there was only one. You find Ajaxian, and start thinking of yourself as super-engineer; the guy who can just as easily write a web UI as a cronjob to perform background data processing.

With the confidence of JavaScript success, you decide to tackle CSS, and then the problems begin. You start from scratch rather than using a library, because after all this is just a display syntax. (Do you also rewrite the math class every time you need a random number generated?) You don’t bother to learn proper semantic HTML, they must be joking! Nattering on about web standards and accessibility. It should be cake.

Nothing works as you expect it to. Your columns won’t line up. You never validated your HTML, and you have a sneaking suspicion that there is an unclosed tag somewhere. You can’t make even the simplest design look right, and you are pretty sure CSS is to blame, rather than your understanding of the technology. This should be just another acronym on your resume, right?

No. Resoundingly, no.

CSS can be predictable, scalable, modular and even object oriented. If it is written correctly, beginners can be productively participating in creating clean, reusable code in 2-3 weeks. But we need to change our approach, understand that the fundamental algorithms for a display language are different than a programing language, and yet borrow everything we can from software engineering, so that we don’t waste time reinventing the wheel.

I visited Yahoo! last week to record a talk I had given at the Front End Summit in October. If you are a designer or an F2E it is essential that you understand the ways in which design choices impact overall site performance. This talk establishes guidelines for High Performance Design including 9 Best Practices.

9 Best Practices

1. Create a component library of smart objects.
2. Use consistent semantic styles.
3. Design modules to be transparent on the inside.
4. Optimize images and sprites.
5. Avoid non-standard browser fonts.
6. Use columns rather than rows.
7. Choose your bling carefully.
8. Be flexible.
9. Learn to love grids.

Web Directions North, Denver, February 2-7

I’ll be speaking more about Design and also CSS best practices at Web Directions North in February where I’ve been invited to give both a Performance Bootcamp Workshop and a CSS Performance for Websites and Web Apps Presentation. I look forward to seeing you there!

80% des problèmes de performance Web se situe au niveau des échanges avec le navigateur et sur le navigateur lui-même : échanges réseau, rendu dans le navigateur, organisation des composants dans une page etc.

Nous aborderons les principales problématiques et les solutions à mettre en œuvre. Forts de l’expérience de l’équipe performance de Yahoo!, à la fin de cette session vous saurez aborder la question des performances Web du point de vue du visiteur et mettre en œuvre les actions correctrices sur vos sites Web.

I’m at Ajax Experience this week with my teammate, Stoyan Stefanov. This morning we did a demo of our new tool SmushIt.com. Smush it allows you to automate image optimization by using the best of open source algorithms to achieve the smallest, high performance images possible.

Smush it comes in different flavors:

• You can upload a bunch of pictures in your browser
• You can provide us with a list of image urls or
• You can get a Firefox Extension to optimize the images found on any web page

Je vais parler (en francais! eek!) avec Eric Daspet de la performance des images pour le web a ParisWeb. Les inscriptions pour Paris Web 2008 sont officiellement ouvertes. Jusqu’au 15 octobre au soir, vous bénéficierez de tarifs réduits. Le conference sera lieu a Paris le 13-15 Novembre. J’attend vous voir bientot alors.

Voila le proposition

Voulez-vous améliorer la vitesse de vos pages web et réduire l’impact écologique et monétaire de votre hébergement ? Voulez-vous faire ceci avec peu de changement de code et en gardant une belle interface graphique ? Cette session va vous apprendre les 7 étapes pour mettre votre site web au régime. Comment perdre des poids que votre site a pris en rajoutant les dernières nouveautés. Et, encore plus important, comment ne pas reprendre ce poids !

The Web Standards Curriculum published by Opera is a great place to start. It will give you the basics of Front-end Engineering from the ground up. The second wave of articles was recently published including a background image and sprites how-to by yours truly.

Time to board a plane, so I can’t tell you more just now, but check it out. It is a great place to get started or brush up on your understanding of web standards.

Website performance is still very much in it’s infancy. Most performance experts come from a traditional engineering background and yet, as Steve Souders showed in his book, the front end is responsible for more than 80% of end user response time.

I was excited when Ed Eliot asked me to be a part of a SXSW web performance experts panel, not only because he and I always have really excellent conversations, but also because of the quality of the people who are on board for the panel. The panelists are uniquely positioned to speak to the concerns of web developers. In addition to being performance experts, we are front-end engineers — as yet a rare combination. In our day to day work, we have experienced the pain points we’ll discuss on the panel.

Who will be on the panel?

Ed Eliot, Stoyan Stefanov, Stuart Colville, and I will join forces to talk about website performance including practical solutions to common problems. These guys are brilliant, and the panel should be really fun.

VOTE FOR US!