A lot of the article is pretty erroneous. While I am an advocate of CSS for a lot of purposes, it is not the end-all of speed enhancements like this article claims it to be. To be blunt, some parts are pure bunk.
1.) The author assumes that size of image files, stylesheets, html pages or any other file size has a relationship to download speed. All files are transferred in packets, which can be of adjustable sizes depending on line quality and errorless successess. Usually (don't quote me on this part) the packet size is 1k + 24 bytes of handshaking information. With that said, a file that is 1k + 1 byte over will take just as long as a file that is 2k dead-on.
2.) The article does not take into account the caching of certain images such as spacer GIFs and erroneously "thinks" that they get downloaded each time they are refrenced in a table when they are used for layout purposes. He eludes that you can have whole slews of these spacers and it will cause extra download time, which is far from the truth.
3.) Browser will read the table in multiple passes, first the structure and any defining attributes such as widths and heights. Widths and heights are not only derived from the table elements itself, but from elements within the table -- such as images. This is why it is good practice to define image dimensions so that the table layout will rendered correctly before the browser makes an actual call for that image.
This is where he "thinks" that the browser makes a second call to download the Html file again. Wrong. The element references for images and the text content are in client-side memory. The text will display instantly at that point. In other words, it does not affect download speed. The images (which would be downloaded anyway) are then called for and displayed at the correct position in the browser.
4.)
CSS generally requires less code than cumbersome tables. Bunk. There is less code in the Html, but it is all transferred to a stylesheet. By his way of thinking, I could logically claim that stylesheets increase download times ... which we all know is bunk too.
One could argue that tables are less cumbersome than CSS postioning. The forums are full of hacks, quirks, and endless hours spent on this crap not rendering proplerly in this browser or that -- then it gets into the old "blame it on IE for not following the rules" line of crap. Hell people, if you want a fluid 3 column layout here is the code. It took me all but 30 seconds to type it in and is guaranteed to work in ALL browsers.
Code:
<table summary="Three column layout guide" align="center">
<tr>
<td width=120>Column One</td>
<td width=510>Column Two</td>
<td width=120>Column Three</td>
</tr>
</table>
5.) CSS can control the order of when content appears on the screen. This is true. But lets think about this for one second. If you put your header into one table -- as soon as it is read and the browser notices the </table> and closes it ... it gets rendered and will appear on the screen. A couple of other tricks of the trade and you can have content appear above the fold while the other is still downloading in the background. This argument is somewhat a moot point in my opinion.
The rest of the article deals in normal specifics that everyone should at least adhere to. CSS for font styles, heading styles, etc. etc. is a good thing. The more of that crap you can take out of there and through into a stylesheet the better.
I would like to point out that they refer to it as a stylesheet, not a positionsheet or a style/positionsheet.
I would also like to personally challenge anyone here to effectively develop a template for this forum that you are looking at right now that is completely done in nothing but CSS positioning. It will also have to render exactly the same in all browsers. Drop me a PM when you have.
Now don't get me wrong. I use CSS as much as possible (for styling) with a minimal amount of positioning when it is not cumbersome to do so. The more I can get into the sheet the better. But I am not going to lose any sweat or sleep at night worrying about why a margin is protuding over the right side of my page and causing some Opera browser to have a horizontal scrollbar appear and start raggin on me about it.
At some point you just got to cut and run with the tables on intricate layouts. Those who are masochistic enough to toil away for three days on some solution, be my guest ... more power to you. But at the end of the day, all you really have done is gave yourself a pat on the back and a "job well done". My solution still looks the same and renders just as quickly as yours will -- and the viewer will not even know the difference.
If you want a good tool to analyze your page with to see some of the possible ways that you can optimize for download speeds, try
http://www.websiteoptimization.com/services/analyze/
This is a sweet tool that will estimate your download times based on latencies due to numbers of packets sent via the transfer. It will identify "objects" (which is what they are) and assist you in building webpages that are built leaner in all facets of the page and its other elements.
Speed up your website download time!!!
Linear Mode
