“I can”t speak to SuperSheep23″s perception of the CSS Working Group, as I don”t know what he proposed nor how he did it. I don”t even know that he talked to the WG—it”s more likely he discussed things on www-style, which is open to all.”

You are right that I voiced myself on www-style and not at the working group meetings. So it isn’t quite fair to say that I brought it up with the working group so much as I brought it up in www-style and members from the working group responded (among others).

I break down CSS’s problems into two categories (fundamental and not).

The first fundamental problem is one of scale:

Even in an all software on a platform renders CSS the same utopia, each destination device requires additional CSS if not a complete CSS re-write. Media queries are what allow you to target these devices, but it only allows you to do the extra work, it doesn’t do that work for you. Even so, it only allows you to account for screen real estate and not device capabilities (mouse, keyboard, touch screen). Creating a usable website for a mouse and keyboard combination is very different than creating a site for a touch screen system is different than creating a web site for a phone or PDA.

So, to me, it doesn’t make sense to specify CSS on the server (versus the client). As you can imagine this went over like a lead balloon. This is a fundamental problem of scale that requires web designers to create far too much content. Given that most sites don’t test in alternate browsers how can we expect them to target alternate devices which may number some day in the hundreds?

—————————————————————–

The second fundamental problem is regards user stylesheets:

The ability to use user stylesheets is effectively an all or nothing thing. Since CSS layouts are constructive (meaning layouts are built from properties on the elements; compare the advanced layout module where elements are assigned to zones), small changes (e.g. font size or line spacing) can ruin a website’s layout. This means that the author of the user stylesheet must overwrite all layout and positioning properties which may also result in a fairly bad experience in practice (ideally as designed it shouldn’t be so bad, but we don’t live in that world).

—————————————————————–

The third fundamental issue is inconsistent grammar:

I’m not the first person, nor will I be the last person to bring this issue up. CSS overloads a number of grammar elements which adds fairly quickly to confusion. The most recent proposed overload is the comma, which will now be used both as a selective fallback (in font) and as an array operator (for multiple backgrounds).

Next there are properties where the order of the values matter (e.g. border-width) and where they do not (e.g. border). This system complicates the parser for what I’m hoping are obvious reason (I’m just trying to keep the length of the post down). This also prevents the addition of properties into compound properties where we cannot use heuristic evaluation (e.g. two properties that can both have a length or both have the keyword black). A prime example of this is the one-off example of font-size and line-height in the compound font property. These one-off issues prevent the kind of abstraction necessary in learning. Each exception must be memorized for CSS to be used which slows down the learning process and slows down non-master usage.

=================================================================

Other problems fall into design issues that aren’t at the core of CSS, but make my life (and others as I would imagine) difficult.

The first issue has to do with requiring me to translate layouts into the box model and attach those properties to elements. Since this produces a constructive layout small changes to the source can have dramatic and undesired effects. The advanced layout module hopes to help with this.

The second issue has to do with lack of portability. Again the advanced layout module seeks to help with this, but I fear it does not do enough. The core layout of a document should exist at the top of the CSS document making it easy to use as a reference. Part of the problems brought up before is the difficulty in reading CSS and determining what a layout will look like given the CSS for it.

The third issue is the inability to place things relative to each other in more interesting ways. There are 9 points of interest on a rectangular object: the four corners, the middle of the four sides and the absolute center. These are easy to calculate, but CSS gives us very little in terms of laying something out in terms of these properties. Centering an object within another object is very unintuitive to me and those I’ve taught, it is the way it is. Having to explain to artists that in order to center something they need to mess with its margins isn’t exactly intuitive for them (admittedly a limited subset of users). CSS though feels like its full of these “mean A, write B” systems. What I’m looking for is the ability to say place object A’s center at the lower right corner of object B.

=================================================================

I’ve got solutions to a great many of these problems, some of which I’ve presented both well and badly to the www-style list.

While my issues with the technology may or may not have merit I would like them discussed. The continual stance that some people have shown (both working group members on www-style and users of the technology) saying that there is nothing wrong with CSS (I am glad you at least admit weaknesses) and that I simply haven’t learned it well enough is insulting to me.

I am not in your “league” Eric Meyer, but my issues with CSS are not based on my lack of understanding. I do understand a great deal of CSS. I’ve used it professionally for a long time now and I keep fairly up to date on the draft recommendations so I know what’s coming up. I also read pretty much everything that comes out of www-style and www-html.

Pretty much everything that runs on computers runs at machine code at some point. This isn’t a very good way to program though we did at one point with punch cards. Then assembler was invented. That evolution started a long chain of new languages that seek to improve and help abstract away the difficulties in our job so we can concentrate on harder and harder problems. CSS is far from done and the fundamental problems I’ve listed will not likely be addressed by the language ever and it may be another decade before I see my non-fundamental, but still important (at least to me) problems addressed. So each time I hear, “you can do it with CSS via this translation of what it is you want to do and that should be enough” I think back to all those programmers who probably thought nobody ever needed anything but assembler.

CSS certainly has its frustrating, hair-tearing, computer-punching moments but it’s a huge improvement on the past… and besides most of the problems really are browser-specific.

I read some of the comments on Dvorak’s post, and saw many of SuperSheep23′s posts. I’m not interesting in rebutting everyone who finds fault with CSS. Some are on target, some are partly on target, and some are way off target.

As for SuperSheep23, he’s right that some layouts are very difficult in CSS. Who could deny this? I would characterize this situation as a set of limitations rather than fundamental flaws, but that might just be semantic quibbling. (To me, “fundamental flaws” are errors so deeply ingrained that no reasonable effort can fix them, whereas limitations can be pushed back or removed without corrupting the core. Others may view these terms differently.) I’ve also found that the vast majority of layouts used on the web are quite achievable with CSS, and most can be accomplished with minimal or no workarounds. Mine is not a universal view, but then, whose is?

Still, are certainly limitations, and I think the CSS Advanced Layout module (which I’ve got to find time to write about) has the potential to push those limitations back quite far.

Some will complain that this isn’t an answer since it will take years for the layout module to be implemented, but that was true of almost everything in CSS at one time or another. Positioning, advanced selectors, pure CSS layouts on commercial sites, any kind of non-table layout at all, print styling… all these things were, at one time, derided as “pie in the sky” and “so far in the future they’ll never get here”. And yet they did.

In fact, I’d love to see the IE team implement the advanced layout module ahead of everyone else (say, in the next major update of IE after IE7) and spur the other browsers to catch up. That would be awesome.

I can’t speak to SuperSheep23′s perception of the CSS Working Group, as I don’t know what he proposed nor how he did it. I don’t even know that he talked to the WG—it’s more likely he discussed things on www-style, which is open to all. But I don’t know, so I can’t comment specifically. I left the WG about two years ago, so if he’s talked to them since then, I missed it.

I think it does a disservice to the entire web community to dismiss problems with CSS, saying if you don”t like it don”t use it.

Maybe you’re referring to someone else, but as for me, I did no such thing. What I basically said was that the cascade is central to Cascading Style Sheets– what with it being in the name and all– and if you don’t like how it operates, then CSS is not for you. I could say the same thing about inheritance, which is probably what Dvorak was really upset about anyway.

People who don”t like the way certain aspects of CSS operate may have ideas (myself included) in how to improve it. Why not hear what we have to say?

I always do. Whether I agree with a given idea is, of course, another matter entirely, but I always listen to and look at people’s ideas for improving CSS. I’ve championed a few of them myself, as it happens.

From somebody that is more newbie than pro, CSS is much like chess. It’s easy to learn, but difficult to truly master. Blaming browser inconsistancies on CSS is much like blaming auto accidents on car manufacturers.

“CSS has fundamental flaws. Seriously. While Dvorak may have missed most of them, they still exist and I’ve brought a few of them up here. The CSS working group has no interest in correcting them as they don’t see anything wrong with the current CSS. Mostly, I think it’s that they’re unwilling to make significant change because they don’t have the time. They admit this. This tells me that the fact that the working group relies heavily on free labor and volunteers is probably a bad design. That’s why I have more hope for Microsoft and Sun as they can pay people to come up with technologies to submit to the ECMA, ISO or ANSI.”

“Again, I tried the proper route taking my concerns to the working group. They deny the existence of a problem. So I’m left with the idea of writing a new spec that now has to leverage CSS out completely. Not something I’m looking forward to, but I start by presenting the idea that there’s something wrong that needs fixing.”

Any rebuttal?