Mixing percent and fixed CSS - css

This is a duplicate from UI.StackExchange.com:
https://ux.stackexchange.com/questions/1004/mixing-percent-and-fixed-css
Should you ever apply percentage and fixed CSS together? Will it cause problems, and if so what kinds?
Does mixing degrade browser render performance?
Will mixing give you weird results on initial load with progressive rendering browsers?
Below is just a dumbed-down example of mixed usage, it could be any mixture. I am not looking for validation of the example. I have heard you should never do what I have in the example below, so I am trying to find out if using CSS in this manner is an issue.
Example mix usage:
<style>
.container
{
width:300px;
}
.cell
{
width:25%;
}
</style>
<table class="container">
<tr>
<td class="cell"><td>
<td class="cell"><td>
<td class="cell"><td>
<td class="cell"><td>
</tr>
</table>

+1 Good question. You may want to have a look at this article: "Fixed-width, liquid, and elastic layout" It goes over fixed width layout (em) and elastic layouts (%), and if you click to go to the next page it looks at 'Elastic-liquid hybrid' - where width: is set one way, with max-width: set the other. I know the article linked to above isn't exactly what you asked, but it's an example of mixed use within a single CSS style.
Edit: After some further reading I did find a quite a few contradictory opinions on the subject. I found several articles that held the idea that "you just can’t mix up pixels and percentages". Though, for the most part, these sites were fairly dated. When I narrowed the search to only articles that have been put up within the past year, things changed a bit. There were still a few opinions against mixing, but they typically didn't explain why, and seemed to of the "I always heard it was a bad idea" variety.
The majority of more recent information that I've found on the topic seems to indicate that mixing percentage with fixed widths is a perfectly acceptable practice, as long as it's done with an understanding of the results.
see:
"Fixed vs. Fluid vs. Elastic Layout: What’s The Right One For You?" (Jun 2009)
"For A Beautiful Web: Changing Man Layout"
...
Full Disclosure: I've been a mixer for many years, without really knowing whether my approach was 'correct.'

This should help clear up when it is ok to mix percent and pixels and when it is not.
Mixing percent and pixel widths wouldn't be a problem when you do it as in your example;
.container
{
width:300px;
}
.cell
{
width:25%;
}
When it becomes a problem is when you reverse the order;
.container
{
width:25%;
}
.cell
{
width:250px;
}
In this case, if the browser window (or the parent of .container) is less than 1000px, 25% on .container will be less than 250px and cause .cell to overflow .container.
It also becomes a problem when you mix percent and pixels in the case of width plus padding;
.container
{
width:300px;
}
.cell
{
width:100%;
padding: 10px;
}
This will cause .cell to have a width of 320px (100% + 10px + 10px), and overflow .container.
Let me know if that helps clear things up.

The way you have it is fine. Each cell calculates to 75px. The only times you have to be careful with percentages is when rounding kicks in.
In your example, if you're container was 303px, each cell would evaluate to 75.66666px and round up to 76px, for a total of 304px which would be larger than the container. That one pixel causes all kinds of trouble.

From the research I have done, it appears how you target your CSS(id,class,universal...etc) is most important in browser render performance.
Efficiently Rendering CSS
Writing Efficient CSS for use in the Mozilla UI
Optimize browser rendering

I can't find any documented evidence with test cases to prove this. Could you point us to where you read about this?
I find mixing the two quite useful and I see it a lot out in the wild on reputed / high traffic sites also.
The only issue that mostly affects older browsers and IE is rounding. Have a read here:
http://ejohn.org/blog/sub-pixel-problems-in-css/
http://agilewebmasters.com/robert/percentage-rounding-and-sub-pixel-perfection/

Related

CSS - calc() on font-size - changing font size based on container size

I have a situation where I have a container element (div) that contains text. This text will sometimes be really large - not paragraph large, but it can potentially exceed the width of the text.
Obviously, in most situations it will just knock parts of the text to the next line, but I wanted to see if calc() can be used on font-size to change the size of the font to make sure it is always fitting within the bounds of the div it is in. Can this be done?
.name { width: 500px; font-size: 64px; }
<span class="name">Sometimes it is short</span>
<span class="name">Sometimes it is going to be really long, and people put long names</span>
I could just limit the number of letters people can use for a name - and to an extent I will, but I am curious to see if this can even be accomplished.
I found this post to do it with Javascript, but that was a long time ago, and I think CSS3 has a lot of new things that may let this be accomplished without any scripting. AutoFill
Here a possible solution:
http://codepen.io/CrocoDillon/pen/fBJxu
p {
margin: 0;
font-size: calc(4vw + 4vh + 2vmin);
/* See:
* http://codepen.io/CrocoDillon/pen/jgmwt
* For some math behind this
*/
}
Font-size is calculated with available size using a function that is not perfect, but may be adjusted to work well in some cases.
Calc is still in it's infancy in terms of support & usefulness. By design it's really just there for doing simple math, like (100% - 20px). It really won't do the math complex enough to make the calculations possible. You are looking for a solution that will size the text based on the amount of space the letters physically take up horizontally (which depends on the letter's individual sizing) and the amount of space available for the containing div.
CSS is abstracted away from the actual element's contents, and it has no way to really discern if something currently "fits" or not. It can layout guidelines for how to handle things when they do or don't fit, but it can't adjust itself according to the content like you want it to. (It's not just you, we've all faced this problem or a similar version of it at some point.)
Check out Chris Coyer's post on what Calc might be handy for: http://css-tricks.com/a-couple-of-use-cases-for-calc/
This is still nearly impossible in CSS only, as the size of each character in different fonts isn't known to us via CSS. There is a jQuery plugin called fitText that handles this sort of thing very nicely.
Just a clarification:
if you use something like:
font-size: -webkit-calc(100% + 30px);
font-size: -calc(100% + 30px);
what this does is add 30px to the 100% default font size, it can't be linked to the container width.
Although, you can do math there like:
font-size: -webkit-calc( 100% * 0.09479166667 - 6.666666669px );
font-size: -calc( 100% * 0.09479166667 - 6.666666669px );
... but it will just calculate it against the 1em.
I recommend you to use text ellipsis
.name{
width: 500px;
font-size: 64px;
text-overflow: ellipsis;
overflow: hidden;
display: block;
}
and if you need to control font-size, do it by #media in different devices
This can be done in pure css using the relatively new vi and vb properties.
https://developer.mozilla.org/en-US/docs/Web/CSS/length
These allow the font-size to be scaled in proportion to the container div's inline (vi) or block (vb) dimensions.
Presently only supported on Safari >= v15.4 and Firefox >= v101
e.g.
.name { font-size: 0.5vi;}
See also: https://caniuse.com/mdn-css_types_length_vi

Advanced CSS Tricks: Capital Initial Letters (Drop Caps) within a CSS3 Multicolumn

Since about a year the multicolumn css3 property matured enjoying support from many browsers. Reason to finally implement it on your website for better design and readability. I thought let's push the envelope and adopt the ancient-but-ever-so-beautiful Drop Caps (=first large initial letter) into the multicolumn. However, certain screen widths break the multicolumn layout in FireFox. What am I doing wrong?
see jsfiddle DEMO
When resizing the window width, you can see the jumping/breaking of the layout in action in IE and Firefox. Below an example. Stuck on whats causing the defects in the multicolumn miss-alignments!?
Sorry for my beardy alter ego selfportrait: I forgot to shave, was staring all day at this problem with no time to tidy up. I promise you though a clean neat shaved portrait back here once this issue is solved!
Above more alignment problems in most screen widths on Internet Explorer 11. Curious Safari and Chrome show the layout faultlessly at all browser screen widths no breakage there.
#multicolumn {
column-count: 3;
-moz-column-count: 3;
-webkit-column-count: 3;
column-gap: 53px;
-moz-column-gap: 53px;
-webkit-column-gap: 53px;
column-rule-color: #EEE;
-moz-column-rule-color: #EEE;
-webkit-column-rule-color: #EEE;
column-rule-style: solid;
-moz-column-rule-style: solid;
-webkit-column-rule-style: solid;
column-rule-width: 1px;
-moz-column-rule-width: 1px;
-webkit-column-rule-width: 1px;
}
#multicolumn p:first-letter{
float:left;
font-weight:normal;
font-size:44px;
margin: 7px 1px 0px 0px;
line-height:27px;
background-color:#AEE;
}
First of all I want to say that the use of the multi column layout module is still not recommendable.
Mainly because of the missing support for the break-before , break-after, break-inside properties, with the exception of IE 10+ and the proprietary -webkit-column-break-* properties (see: CSS3 Multiple column layout).
(You also may want to take a look at my answer to this SO question: IE (11) improper handling of CSS multi-columns?)
Additionally you have to remember, that there is a so called "multi-column pseudo-algorithm", which seems to be confused by your :first-letter selector.
You can avoid this problem by using a span element with a class attribute for the drop caps instead.
But as the first letters are larger in size than the rest of the text, there arises another problem.
It may happen that a (single) line of text of the beginning of a paragraph with a drop cap may fit to the previous column, whereas the drop cap (which is about twice as high as the normal text) may not.
To avoid this unwanted behaviour you have to use another span element, which includes at least more text than that may fit into a single line (of text)!
And giving these span elements a display: inline-block; solves this problem.
Just a word about Amir5000 answer: Though my proposed solution also needs some extra span elements, it does not use "purely presentational markup" which may also produce unwanted empty lines.
But as said at the beginning, using multi-column is at least very "tricky" and very difficult to get predicted results across browsers and/ or different viewport widths.
So here is my proposed "solution": DEMO
The cause of the issue was the float:lefton the #multicolumn p:first-letterif you take that out you will see it no longer has that issue; However you don't have the same format you wanted with the first letter. So I created a JSFIDDLE where I added
#multicolumn p {
float: left;
}
and added a width for the #multicolumn container and centered it as you can see.
Hope that resolves the issue for you.
-------Update---------
So after much time trying to get it to flow as intended I was able to come up with a work around that is pretty simple, if you take a look now at the updated FIDDLE
I added an empty span in between paragraphs to clear the float and also added a media query so it looks nicer on smaller screens
This was the simplest way I could come up with to solve your issue hope that helps!!
I am not sure about the the column issue. You should make a fiddle for it so that we can help you faster. As far as the first cap issue. This is pretty tricky, what are your constraints? Can you hardcode it look right or do you have to do it dynamically?
I have posted a hardcoded solution here. It is basically just using
:before
http://jsfiddle.net/emersive/bdAWQ/1/
I have Chenges below css
p { float:left;}
#multicolumn { line-height: 20px; } /*need for IE browser*/
Demos

CSS beginner questions

I'm about to start designing my site that I coded in php. It's a rather complex site...
Obviously, I want to make it suitable for all browsers and devices.
Should I code the CSS using fixed pixels for normal computer dimensions and when I'm done with that make another stylesheet for other devices such as iphone and ipad?
Or should I attempt to make an adaptive site that works with all resolutions? In this case, should I look into CSS frameworks? Any recommendations?
Here's some tips for designing a web page that works on ALL screen sizes. I'll delve a bit into media queries and other things too.
On web pages I've done in the past they usually have existing graphics (usually banners/headers, etc.). so I usually design the web site using a maximum width for example, if the banner image that already exists is 900px wide I'm going to make the page 900px wide, more specifically the content wrapper will be 900px wide and it's contents will sum to 100%.
Keep everything in a global content wrapper. This is the easiest way to ensure the page size stays 100% consistent.
The sum that makes up the wrapper should always amount to 100%. Fill the wrapper, but it will never go outside of it. You may have numerous columns or divisions, which you can size at any percentage up to a sum of 100. (like, col1 width="33%", col2 width="33%", col3 width="34%" = 100%)
Try to pick images that work for ALL screen sizes. There is no built-in way to swap to mobile images when working with responsive web design (there are javascript solutions), but pick an image that looks good when scaled down.
Avoid scaling an image up. This is why i tend to make the max page size the same size as a header image (if there is one). If the page has no images, 100% liquid layouts are very good too.
If the content wrapper is fixed, turn it into a percentage with a media query :D example below.
I'm going to write up a basic example of a MAX width 900px page with columns that works on any site mobile/desktop.
default-style.css
img
{
max-width: 100%;
height: auto;
width: auto; /* ie8 */
border: none;
}
.content_wrap
{
width: 900px;
margin-left: auto;
margin-right: auto;
height: auto !important;
}
/* Columns */
#column-wrap
{
float: left;
width: 100%;
}
#column-left
{
width: 30%;
float: left;
}
#column-middle
{
width: 35%;
float: left;
}
#column-right
{
width: 35%;
float: left;
}
default-queries.css
/* Turn it into a liquid layout at our max-width of 900px */
#media screen and (max-width: 900px)
{
.content_wrap
{
width: 100%;
}
}
/* most smart phones */
/* This query will turn all 3 columns into 1 column at 540px */
#media screen and (max-width: 540px)
{
#column-wrap
{
position: relative;
clear: both;
}
#column-wrap div
{
margin-bottom: 10px;
float:none; clear: both;
}
#column-left
{
width: 100%;
}
#column-middle
{
width: 100%;
}
#column-right
{
width: 100%;
}
}
Now for the html :D
index.html
<!DOCTYPE html>
<html>
<head>
<link href="default-styles.css" rel="stylesheet" />
<link href="default-queries.css" rel="stylesheet" />
</head>
<body>
<div class="content_wrap">
<div id="column-wrap">
<div id="column-left">
<p>HELLO I"M LEFT COLUMN</p>
</div>
<div id="column-middle">
<p>HELLO IM MIDDLE</p>
</div>
<div id="column-right">
<p>RIGHT</p>
</div>
</div></div>
</body>
</html>
So this is a most primitive example of a responsive web site. Using the class content_wrapper at a fixed width of 900px (because the biggest image im using may be 900px). But when the screen is smaller than 900px, we switch from fixed to liquid via media query switching to a liquid width will allow our images to scale down, our text to wrap, and a whole bunch of other great things!
But, there's more to it than that. That's all we need to do if the page only has linear content, but most pages have different columns to make the most of large screen sizes. This is where media queries REALLY come in handy. If you study the columns, their sum takes up 100% of the content_wrap at all times. And they will scale down for a small website, but they will be really really tiny. This is why I make another media query for smaller screens and eliminate the multi-column layout. Simply clearing the floats and making each column have a width of 100%, will allow for a much more efficient mobile layout.
Anyway, really long answer but hopefully my views/opinions on how to properly design a website with CSS/HTML (php, ASP.net, MVC, etc, can all be thrown in there too) will get you started! Design with a wrapper, media queries, no upscaled images, and images that look good scaled down are really good practices IMO ;)
I like fixed width pages! They are more universal and prevent over-scaled images Media queries in CSS3 take care of responsiveness ;)
I usually DON'T do mobile-first designs, but that's just my personal preference. I see no advantage to one way over the other
As stated in another answer, for non-CSS3 browsers response.js works very well
As usual, it depends. If you think your site is going to be accessed by many devices then you probably want to look at responsive design: http://johnpolacek.github.com/scrolldeck.js/decks/responsive/ and http://www.alistapart.com/articles/responsive-web-design/ for more details.
The recommended approach is to take a 'mobile first' approach to the styling, so it works well with little markup for mobile devices and then progressively enhance for other devices with larger resolutions.
You may want to use frameworks or hand-code it - that's something of a personal choice.
If you have significantly different needs for different devices, you may want to create entirely different sites for mobile, similar to the current BBC news for instance, although it's worth noting that many sites including BBC are moving to responsive techniques, so this is a trend that is being adopted more and more.
It's also worth considering which browsers your audience will be using. As you reach back into older IE versions especially you'll need to look at javascript techniques such as the respond.js library to allow media queries to work.
The two aren't mutually distinctive.
The two approaches are generally:
Multiple stylesheets, media queries decide which one to load.
Use fluid measurements (percentages) so the site is fluid.
In reality, you often end up using a bit of both.
There are many issues with trying to do this, one particularly being that older IE doesn't support media queries, and renders everything all the time.
Frameworks are good, but you may find them restrictive. There are many around, and they all basically do the same thing, so just google Responsive CSS framework and see what you find.
I'd also recommend looking into SASS at this point - I use it so I can auto generate an IE stylesheet that has the 960px width version with no media queries, whilst generating another sheet with all the queries in.
Already been answered, I see, but I think what I've got to say is relevant...
Most sites do both a desktop site and a mobile site. Desktop sites are almost always fixed width, with a maximum width at about 1000px. This is starting to change, though, you will see more adaptive sites than in the past. Mobile sites need to have an adaptive width, because of all the crazy different resolutions on all the various phone models out there. I recommend that the smallest width the mobile site should work at is about 320px, because most modern phones are at least that width.
As for mobile detection, take a look at PHP Mobile Detect, it's what I use, and it works pretty well. You just throw in
<?PHP
if ($detect->isMobile()) {
header('Location: /path/to/mobile/site');
} else {
header('Location: /path/to/full/site');
}
?>
at the very top of the first page. You should use a seperate splash page that redirects to something like index2.htm so that users can choose to view the desktop site if they want. It saves you a lot of trouble trying to figure out how to set cookies...
I recommend making an adaptive or responsive that can maintain its look across multiple device resolutios. With that being said CSS framework will be your option so your html file is not filled with your topic content and style code.

Is there a CSS workaround for Firefox' bug: inline-block + first-letter with changed size

It's better to see a bug for yourself in Firefox: http://jsfiddle.net/kizu/btdVd/
The picture, showing the bug:
And the bug filled in 2007 on bugzilla.
The bug appears when you're adding ::first-letter pseudo-element with display: inline-block, and then change the font-size of this first-letter.
More letters in a word after the first: more extra space added (or subtracted — if the font-size is lesser than block's).
Adding float: left to the first-letter inverts the bug: with bigger font-size the width of inline-block shrinks.
So, the question: is there any CSS-only workaround for this bug? It's somewhat killing me.
I've found that triggering reflow on the whole page (or any block with a problem) fixes the problem, so I've found a way to trigger it on every such block with one-time CSS animation: http://jsfiddle.net/kizu/btdVd/23/
Still, while this fix have no downsides in rendering, it have some other ones:
it would work only for Fx5+ (that supports animations);
it still flashes the original bug for a few ms, so it's maybe somewhat blinky.
So, it's not an ideal solution, but would somewhat help when Fx4- would be outdated. Of course, you can trigger such fix onload with JS, but it's not that nice.
I don't think there's a good solution.
I have come up with a flaky solution for you though:
.test:first-letter {
font-size: 2em;
letter-spacing: -0.225em;
}
Add the letter-spacing style to the :first-letter selector in your Fiddle, and you'll find the blocks go back to roughly the right size.
Explanation:
Basically, the bug is being caused by the whole block taking its size from the font specified in the first-letter.
What I'm doing here with the letter-spacing is trying to adjust the size of this font, without affecting it's physical appearance. Adjusting the letter spacing in this way in normal text would result in the letters overlapping each other by .225 of a character width on either side.
I was initially hoping that a value of -0.25 would be sufficient -- ie a quarter of a character on each side would reduce the width of each character by half, which would be what we want because the first letter is font-size:2em, so it's twice as big.
However, the calculation isn't quite as clean as that, because the first and last characters in the string would only be overlapped on one side each, and because the first letter does in fact want to be double width, even if the rest of the characters don't.
All of this means that the exact letter-spacing value required to counter-act the bug will vary depending on how long the text, as well as the font sizes of the original text and the first letter. This is why I had to experiment a bit with the value of the letter spacing to get it working, and also explains why I couldn't get quite a perfect fit on all the text rows in your Fiddle. I would have needed a slightly different value for each block.
The value of -0.225 seems to be about the closest I can get to it being right for all your examples, but in practice you'll need to adjust it to suit your site.
Don't forget also that this is a Firefox bug, and therefore you'll need to write it in as a browser-specific hack of some sort. And be careful to keep an eye on the Firefox bug report you linked; when it does get fixed, you'll need to work out a way to keep your hack in place for users of old versions, but remove it for users with the fix.
[EDIT]
The only other working solution I've come up with is simply not to use the features which trigger the bug. ie drop the :first-letter selector, and use a separate <span> for the first letter of your text if you want to style it differently.
This is the obvious answer really, and would of course solve the problem (and would also mean that your styled first letter works in older browsers), but it would not be ideal from a semantic perspective, and more importantly doesn't actually answer the question, which is why I didn't offer it as a solution in my original answer.
I have been trying to find alternative work around for the bug as well, but the options are limited, and nothing I've tried has given as good results as my initial suggestion.
I tried a hack of making the :first-letter invisible, and using :before to display the big leading capital letter. However, this didn't work for me. I didn't linger on it too long so you may be able to get it working, but there is a problem with it in that you'd have to define the leading letter in your CSS, which wouldn't be ideal.
Another possible solution is to use the CSS font-stretch: condensed; property on the :first-letter. This would reduce the width of the first letter back to 1em, and thus resolve the width issue of the rest of the text. The down sides of this, however, are that firstly it would make the leading letter look squashed, which is probably not what you want, and secondly this style only works for fonts which support the condensed property. It turns out that this isn't well supported by the standard fonts, so may not be workable for you.
In the end, the original letter-spacing solution is still the only way I've found to really work around the bug.
This bug still exists, but some of the fixes don't work anymore. Even after triggering a reflow with an animation, the inline-block returned to the same size for me. I couldn't use the letter-spacing trick because I am already using it on the first letter, that is what is causing the problem for me. I solved the problem by adding this to the CSS for the affected selector:
-moz-padding-end: *number of pixels to compensate*;
At the moment, moz-padding-end seems to be specific to Firefox, and it can add or remove width to the end of the inline-block. Because this is a Firefox specific bug, that did the trick for me.
I know this thread is quite old now, but apparently this bug has not been fixed yet.
Using animation does work but there is a noticeable FOUT (Flash Of Unstyled Text). I was able to work around the problem by wrapping the first-letter in a span. This does work around both the sizing issue and the FOUT, it does add extra elements to the markup, so it depends on the needs of your site/application if this is the best fit.
.test {
background: rgba(0,0,0,0.15); /* For visualisation */
display: inline-block;
}
.test:first-letter {
font-size: 2em;
}
.test2:first-letter {
float: left;
}
.test3:first-letter {
font-size: .5em;
}
<h1>Inline-block with bigger first-letter</h1>
<span class="test">Broken</span>
<br><br>
<span class="test"><span>F</span>ixed</span>
<h1>+ floating to first-letter</h1>
<span class="test test2">Broken</span>
<br><br>
<span class="test test2"><span>F</span>ixed</span>
<h1>small size for first-letter</h1>
<span class="test test3">Broken</span>
<br><br>
<span class="test test3"><span>F</span>ixed</span>
<h1>small size, floating first-letter</h1>
<span class="test test2 test3">Broken</span>
<br><br>
<span class="test test2 test3"><span>F</span>ixed</span>
As of 2023, this is still happening in Firefox.
This is my solution using SASS, but you can see how to make it bare CSS:
txt-brand {
display:inline-block;
}
///Firefox only ///
#-moz-document url-prefix() {
margin-right: .1em;
white-space: nowrap;
&::after {
content: '\00a0';
}
}
}
.txt-brand::first-letter {
letter-spacing: -.11em;
}

Min-width in MSIE 6

What is the definitive way to mimic the CSS property min-width in Internet Explorer 6? Is it better not to try?
foo { min-width: 100px } // for everyone
* html foo { width: 100px } // just for IE
(or serve a separate stylesheet to IE using conditional comments)
You could use an expression (as suggested by HBoss), but if you are worried about performance then the best way to do this is to add a shim inside the element you want to apply a min-width to.
<div id="container">
The "shim" div will hold the container div open to at least 500px!
You should be able to put it anywhere in the container div.
<div class="shim"> </div>
</div>
#container .shim {
width: 500px;
height: 0;
line-height: 0;
}
This requires a little non-semantic markup but is a truly cross-browser solution and doesn't require the overhead of using an expression.
This article on CSS Play, by Stu Nicholls, shows the different methods for achieving min-width in IE, in all modes (Quirks, etc) and even for IE/Mac.
I've fiddled with every answer given here in the past month. And after playing with Pretaul's method (Min-width in MSIE 6), it seems to be the best alternative to min-width. No hacks or anything, just straight up compliant CSS code which takes 30 seconds to implement.
From Googling around, expressions seem to be the most popular. For me anyways, ittended to randomly lock up my browser (both IE and FF).
I dunno, I had some success with:
min-width: 193px;
width:auto !important;
_width: 193px; /* IE6 hack */
A combination of dustin diaz' min-height fast hack & How do I specify in HTML or CSS the absolute minimum width of a table cell
do your css tag as _Width: 500px or whatever.
This works pretty well...
div.container {
min-width: 760px;
width:expression(document.body.clientWidth < 760? "760px": "auto" );
}
Min-height fast hack works for me (also works for width)
The shim example is fine for forcing the browser to show a horizontal scroll bar when the container gets to a certain size but you'll notice that the content in the container will still be resized as the window gets smaller. I imagine that this is not the overall goal when trying to achieve minimum width in IE 6.
Incomplete min-width technique http://www.mediafire.com/imgbnc.php/260264acec99b5aba3e77c1c4cdc54e94g.jpg
Furthermore, the use of expressions and other crazy CSS hacks just isn't good practice. They are unsafe and unclean. This article explains the caveats of CSS hacks and why they should be avoided altogether.
I personally consider scaryjeff's post to be the best advice for achieving true min-width in IE6 and as an experienced CSS layout developer I've yet to find a better solution that is as applicable to problems of this kind.
This article on CSS Play, by Stu Nicholls, shows the different methods for achieving min-width in IE, in all modes (Quirks, etc) and even for IE/Mac.
I've provided an answer to a similar question that details the use of this technique to correctly achieve min-width. It can be viewed here:
CSS: Two 50% fluid columns not respecting min width
The technique is simple, valid CSS that can be used in almost any situation. Applied to the shim example above it results in what I consider to be correct min-width functionality.
Correct min-width technique http://www.mediafire.com/imgbnc.php/a67b2820bfbd6a5b588bea23c4c0462f4g.jpg
Single line button
button{
background-color:#069;
float:left;
min-width:200px;
width:auto !important;
width:200px;
white-space: nowrap}
Use conditional comments to reference and MSIE 6 specific style sheet, then create CSS as below.
Compliant browsers will use:
min-width: 660px;
Then MSIE 6 will use:
width: expression((document.body.clientWidth < 659)? "660px" : "auto");

Resources