DOM Traversal

re: DOM Traversal

wetran — Tue, 10 Aug 2010 06:19:15 GMT

@Mitch 74,

Your personal programming habits do NOT mean they should be obeyed by everyone.

"one VAR statement is allowed by function or program body: don't do 'var i=something;var j=else;' but do 'var i=something,j=else;'. Note that you can declare a variable without setting it, like this: 'var i=something,j=else,x;'

If declaring variables at the start of a function is so difficult, why do they do it in Delphi and in C (and all its descendants), I wonder. And while ECMAscript indeed allows multiple VAR statements, it's good practice (and a performance saver) to allocate all your variables at the start: the parser and evaluator will not need to increase the "stack's" size (or more probably, have to re-work the index) more than necessary."

Please, we are not in the 1980's anymore. I've been programming since the early 1980's, and yes in the early days performance and resource-saving WAS critical so people usually recommends those kind of "tricks", but now we are in the 21st century, and currently we teach people program readability > performance and resource-saving, unless it's some mission-critical low-level programming for stuffs like SoC and extremely limit resources. And one variable declaration for a whole program body is a big NO in this case. People prefer clearer and readable code much more than putting all kinds of important and temporary variables in one line at the start of a program. The performance gain of putting all variable declaration at the start in one line is negligible given today's hardware resource and interpreter/compiler/JIT compiler optimization technologies.

You can go look at any of those large websites and javascript frameworks out there, like Google's web apps, Apple's sites, Prototype framework, etc. etc. and you'll find them using one VAR for one important variable declaration, for better code readability over negligible performance gains. For these couple decades, stuffs like "var i=something,j=else,x;" is the bad practice. Every program guide tells you to do "for (int i=0;i<something;i++) for (int j=0;j<else;j++) { ... }" nowadays, NOT "int a,b,c,i,j,k,...; ... after 100 lines of codes ... for (i=0;i<something;i++) for (j=0;j<else;j++) { ... }", although the latter theoretically helps performance, it's a big NO-NO in this day and age.

Sames goes for the "!=" vs. "!==", it's like saying people should do "if (0==i)" instead of "if (i==0)", that's just some recommendations that theoretically improves productivity, but in reality it's all down to personal preferences, the extra time costs to consciously typing the extra "=" everywhere when it's not exactly needed, may far surpass the potential and theoretical productivity improvement.

re: DOM Traversal

anti-whiner — Tue, 03 Aug 2010 20:47:58 GMT

Writing examples that are understandable is to be commended. And ALT is a supported attribute on OBJECT, AREA, APPLET and INPUT in addition to IMG.

re: DOM Traversal

ryan — Tue, 03 Aug 2010 20:24:05 GMT

Why does MSFT always find odd examples to use?

If you are testing for the alt attribute, then you are looking for images... those images are already available in a DOM-L0 collection;

document.images

thus you can save piles on the traversal with:

for(var i=0,iL=document.images.length;i<iL;i++){

//do tests...

}

or even document.getElementsByTagName('img');

You are surely not suggesting that developers should ignore the specs and add alt attributes to all kinds of other elements are you?

re: DOM Traversal

Dave — Tue, 03 Aug 2010 14:27:20 GMT

> I’ve already found these features to be a great help in my own coding ...

How so? Since a web site needs to deal with downlevel browsers, there will need to be a completely different code path for the two cases. That's why we have frameworks like jQuery and Prototype trying to hide those differences. So I can see how frameworks might find these DOM traversal features useful, but if you're using them bareback it seems like you'll end up writing a lot of framework-like wrappers around them.

Given that querySelectorAll exists, a NodeIterator or TreeWalker don't seem that handy. I wouldn't be surprised to find, for example, that document.querySelectorAll("[alt*=space], [alt*=flight]") is much faster than your example above, since there's no Javascript involved in the node inspection or tree walk.

re: DOM Traversal

BrianMB — Mon, 02 Aug 2010 21:17:12 GMT

- CSS Transitions

- CSS 2D Transforms

- CSS Border Image

Begging on both knees, IE Team. Please make it happen!

re: DOM Traversal

greenunicorns — Mon, 02 Aug 2010 20:23:41 GMT

Huh? Oh, was he accusing me of that? I assumed that he was just adamantly declaring that HE did not live in his mother's basement. I completely missed that. Either way, where I live is irrelevant, and my credibility is not the issue. For the sake of argument, let's go ahead and say I live in TEN mother's basements... regardless of this, you either agree that "same markup" is a silly promise that a single browser maker can't deliver on, or you ascribe to some other belief. I, anonymous Internet commenter, and my illustrious reputation, neither add weight to nor subtract weight from my simple argument.

re: DOM Traversal

rofl... — Mon, 02 Aug 2010 20:09:57 GMT

Kiddo, defending yourself as a "SouthPark-watcher" isn't going to win you a lot of credibility when someone accuses you of living in your parents' basement.

re: DOM Traversal

greenunicorns — Mon, 02 Aug 2010 19:49:11 GMT

@NotLivingInMomsBasement

Ah HA! It was a South Park reference to a Star Wars metaphor, which flags me as EXACTLY one of the hundreds of millions of people they're trying to reach, and your failure to publicly recognize that flags you as somebody who has no right to be flagging me as a... em... I don't know.

Anyway, yes, they can attempt to adhere to other browsers' implementations IF there is a consensus, but they can do nothing for the cause of "same markup" if Opera, Firefox, and Safari are all handling something in different ways. The best they can manage in such a case is "Slightly Less Different Markup". Don't get me wrong, I applaud the efforts in the direction of coming back from Obscure Microsoft Island and joining the mainland where everybody else is, but it just sounds dumb for a browser team to be boasting that they are making their browser so that your HTML will work the same way in all browsers. It just doesn't make sense. Did I mention Chewbacca?

re: DOM Traversal

NotLivingInMomsBasement — Mon, 02 Aug 2010 16:34:41 GMT

@unicorn: They don't need to control what the other browsers do-- they can simply test them. If there is consensus (spec or no) then they can match their behavior with IE9.

Your choice of a Star Wars metaphor clearly flags you as NOT the sort of reader they're optimizing for with the "same markup" slogan-- they're trying to reach the hundreds of millions of more normal people who understand "same markup" == "good"... good for business, good for users, good for the web.

re: DOM Traversal

greenunicorns — Mon, 02 Aug 2010 14:48:14 GMT

@Jim

"Please stop with the "same markup" slogan. JavaScript is not markup and it makes you look utterly clueless when you refer to it as such."

YES. Not to mention the fact that you can't possibly engineer a single browser in a way that ensures that developers can write markup that will be rendered the same by the browsers you DON'T control.

This was obviously the idea of some less-than-technical manager, and I'm sure the hard-working IE9 team is silently bearing the painful stupidity of this slogan. It makes about as much sense as Chewbacca living on Endor.