Conflicting Absolute Positions

by Rob Swan

77 Reader Comments

Back to the Article
  1. Why is it that even on a site devoted to web design, none of the web developers know very much about the languages and APIs they’re using?

    Perhaps because many of us are too busy making websites to study the specs comprehensively? Web design is one of those fields where you learn something when you need it, so overlooking what may seem like an obvious thing is quite easy. Besides, many people generally thought this wasn’t possible (probably through word and mouth) and felt no need to confirm it for themselves (why would they?).

    Copy & paste the code below to embed this comment.
  2. Robert, a good referencing example indeed.

    But I’ll only use this for a client on a very low budget or if the client requires a very quick reference guide.

    Else for a ‘help’ feature, I’d rather advice them to implement a ‘smarter’ system to help produce a solution for a problem. :)

    Still I really appreciate this tutorial. A good reference for me to keep for future. Regards.

    Copy & paste the code below to embed this comment.
  3. Check out this method:
    http://web.hc.lv/kods/css/raksti/2-kolonnu-majaslapas-izkartojums-ar-css/
    Main content goes first in source, which is better for SEO, usability and accessibility.

    Copy & paste the code below to embed this comment.
  4. this goes a bit pair shaped in safari for me—safari 1.3.2 (v312.6)

    it’s the strangest bug—i’ve never seen it before, maybe it’s just a bug on my system not one everyone is likely to see? no idea:

    clicking on the “finished layout” link gives the page, but the background, outside of the three main panels, remains what was on the preceding page. and changing the size of the window results in all sorts of pixel distortion and noise in the areas outside of the three main panels. quite a strange, changing as you change the size of the window, mess.

    Copy & paste the code below to embed this comment.
  5. I have a recurring dream where I write an HTML page, apply CSS rules, and the page renders correctly no matter what the browser. Then I get chased down a corridor by a giant blue lowercase e and I wake up screaming.

    Copy & paste the code below to embed this comment.
  6. Just wanted to post a thank you, this simplified a dynamic resize I had in a project, and also opened my eyes to a sweet new way of fill-up-page block structuring.

    Copy & paste the code below to embed this comment.
  7. I particularly like the clarity of your thinking – taking things back to first principles. It is an object lesson in keeping things simple.

    Your solution is just what I’ll need to implement some redesigns of old frame-based websites that I’ve been putting off.

    Copy & paste the code below to embed this comment.
  8. Thanks.  This was an interesting and helpful little article, Rob.  Nice job!

    Copy & paste the code below to embed this comment.
  9. Firstly, I have to say this is a fantastic solution; it never would have occurred to me to set more than two positions on an element. Very well-written, too, thankyou very much.

    My first point is; No Need for Dynamic Properties
    This is related to Anatoly Litsov’s point above (dynamic expressions can get very expensive since they constantly update). My solution is simply to use a few lines of javascript to do the same thing – get element by ID, do the math with document.body.clientHeight/Width, and have the whole thing wrapped up in a window.onresize event handler, and all of that within the existing conditional comment.

    You mentioned several problems with javascript solutions, inluding problems with cross-browser event handlers; not a problem here. Your method (as you noted)is essentially serving javascript disguised as CSS, using a conditional comment, so why not serve up regular javascript? Critically, this wouldn’t have as much of a performance hit as dynamic expresions; and in my eyes, commits no more ‘crimes’ than the current method.

    Secondly, An attempt to fix the Opera bug

    That bug annoys me a lot, and I would really rather avoid it if possible (aiming for maximum possible backwards compatibility). So, I dreamt up a possible solution with just one HTML change; a container div around the sidebar, with the following CSS (note, my sidebar is on the right, with HTML source order adjusted accordingly)

    #sideContainer {
    position:absolute;
    top:0px;
    right:0px;
    height:100%;
    width:220px;
    }

    #side {
    padding: 0;
    margin: 0;
    position: absolute;
    top: 100px;
    right: 20px;
    bottom: 20px;
    left:0px;
    overflow: auto;
    }

    The idea being that the container div handles the 200+padding width, and the 100% height of the body (html needs 100% height too), and the sidebar is ‘pinned’ with conflicting absolutes within this box. This works in Firefox 2, Safari 3 winbeta, IE6…but in Opera 8, the behaviour is the same as before! Applying a background color to the container shows that it is, as one would expect, stretching all the way to the bottom of the window, including when the window resizes; but the sidebar within it refuses to pin to it’s bottom.

    I’m stumped. I’m posting in the hope someone will spot a flaw, or be inspired to figure out some other way around this tedious bug.

    Copy & paste the code below to embed this comment.
  10. There are 2 issues that the article does not address that will have people scratching their heads trying to figure out why things are not working. Both are covered in my article “Frames wihtout frames”

      http://www.fu2k.org/alex/css/frames

    1. When Microsoft released XP2, it upgraded IE6 so that javascript expressions within CSS are not executed if, and only if, the zone which a page resides within’s security is set to ‘High’. Even if javascript is enabled. By default that is usually set to ‘medium’, but in many corporate/institutional settings that’s not the case.

    2. The expression that Rob uses will only work if IE6 is in quirks mode (the example document has an xml prolog which does the necessary quirkifying). See the “Standards Mode compatibility for IE 6” section in my article,  for an extended expression that will work either way.

    Copy & paste the code below to embed this comment.
  11. @John Campell, others—

    I thought I had a way to fix the 100% & float problems inside the absolutely positioned elements in IE—add position: relative; to the .container divs.

    However, that didn’t do it.  But I was unable to find a problem with floats in IE, anyway, just with width: 100%.

    Copy & paste the code below to embed this comment.
  12. Why stress about Opera 8? “I love Opera”:http://www.alanhogan.com/conversations/i-love-opera-the-browser , but don’t most users (and there are not many) use the latest version (9), in which there is no bug?  And as noted in the article, it only shows on resize. That’s a user-initiated problem that is fixable by reloading, something an Opera user would likely be intelligent enough to try.

    Copy & paste the code below to embed this comment.
  13. <div id=“right”> in the example shoud be <div id=“MAIN”>

    Copy & paste the code below to embed this comment.
  14. I didn’t realise that so many people were vague on positioning so its good that your article brings this to attention.

    As a matter of interest you might like to know that the original example can easily be done without using expressions (and in standards mode)and I have converted one of my very (very old) demos to look like the authors.

    http://www.pmob.co.uk/temp/2scroll2.htm

    or an even older similar example.
    http://www.pmob.co.uk/temp/3scrolling2.html

    Instead of using expressions to subtract pixel heights from 100% you simply apply top and bottom padding to the html element and then IE interprets 100% height of the body from inside the padding thus giving you your gaps at the top and bottom of the page.

    Of course it only works in 100% height scenarios and so is of limited use. I often find that these types of layouts (along with forced fixed positioning examples) only work well in certain situations but not for everything as there are quite often scrollbar issues.

    I also notice this quote early in the article which is a little misleading:


    “Technically you can use both a right and a left property and leave the width as auto. But unfortunately IE does not support this”¦”?—css-discuss

    That isn’t quite true and IE does support the properties left and right at the same time and you would soon see this if you tried it out.

    #main {
    position: absolute;
    top: 100px;
    left: 240px;
    right: 120px;
    background:red
    }

    The misconception arises because the background appears to only enclose the content but if there is enough content then it stretches correctly to the positions defined. In fact this can be cured by giving the inner content “haslayout” and then it works as it should.

    Its only top and bottom positions that IE doesn’t understand when used together.

     

     

    Copy & paste the code below to embed this comment.
  15. #44: the bug you describe does surely exist. I can confirm both the background from the preceding page and pixel distortion on window resize.

    I am really interested in a solution to this problem. I would like to adopt this approach, but my company officially supports Safari 1.3.2 :=(

    Copy & paste the code below to embed this comment.
  16. Sorry for the huge text in 55, it looked fine in the preview..

    Scrolling does not work in Safari 1.3.2 – while the scroll bar is moving, the content is fixed, which means that you can never see any content below.

    Copy & paste the code below to embed this comment.
  17. Excellently post and easy to understand.

    This post has saved a lot of time. Thanks for your time to explain everything.

    Copy & paste the code below to embed this comment.
  18. The fix is to add height:100% to html tag.

    Copy & paste the code below to embed this comment.
  19. Has anyone tried using this technique to create a 2- or 3-column layout that goes beyond the bottom of the window and where both or all of the columns stretch to the same height as a “height-defining” column?  I believe I’ve figured out how to do this in all browsers (no, I didn’t use faux columns or funky paddings or margins).  I was wondering, Rob, if you’ve seen this before and if not, if you’d be interested in hearing my method.  I read an article that said this would be the “holy grail” of CSS web design.

    Copy & paste the code below to embed this comment.
  20. I just found Matthew Levine’s article.

    Copy & paste the code below to embed this comment.
  21. I use this technique or variants very heavily, thanks to the frames without frames article published (and cited) a while back.

    Be forewarned: VERY rarely, and usually inconsistently, Expressions can cause the IE browser to hang in an infinite loop. What appears to happen is the dimension gets calculated which causes the expression to trigger again, ad infinitum. There are some discussions of this, but I don’t have the links.

    For me, it was the width being adjusted. In the end I had to pull it from the CSS,and put it in a javascript embedded in the footer of the page. I didn’t like doing so, because CSS is where it should be, but I had to. So people should be aware of this. While I was at it, I fixed the opera issues in JS too (window.opera is a good way to object sniff for opera)

    Copy & paste the code below to embed this comment.
  22. Hi, i was looking for the same layout for some time ago, found the same solution with the absolute positioning of each corner, and ran in to the same problems in IE6 og earlier versions.. After a bid of work i actualy manged to find af solution that works alle the way down to IE5 in pure CSS without the need for Javascript. – Will put it online and post a link if anyone should be interessetet :-)

    Copy & paste the code below to embed this comment.
  23. Hi.. Haven’t had time to write an explanation, but the code should be pretty strait forward :) (There is a small problem with the positioning og the right side scrollbar in IE6 that i haven’t been able to solve yet, but other than that i works as a charm as far as i can se :)

    http://www.einventions.dk/kongknabe/frameset_layout.html

    Copy & paste the code below to embed this comment.
  24. Rasmus – see my post 54 :)

    Copy & paste the code below to embed this comment.
  25. Sorrey.. Missed your post :-)

    Copy & paste the code below to embed this comment.
  26. Does anyone know of a way that I could achieve a similar effect using the “em” size unit?
    I would like my header to be able to fully encapsulate its text, regardless of the text size the user has chosen.

    Copy & paste the code below to embed this comment.
  27. I noticed that plenty have reported the problem with absolutely positions containers that incorrectly size containing DIVs with 100% widths.

    I found that it no scrollbar appears in IE7/Win, FF2/Win, O9/Win, NN9/Win, S3/Win.

    The problem I’m experiencing is that I cannot set the height of DIVs to 100% in IE7/Win. It works in all the above mentioned browsers.

    Have anyone got a solutions for this problem?

    Copy & paste the code below to embed this comment.
  28. Thanks for taking the time to write this great, clear article.

    I’ve done this layout using pure CSS, but using individual BORDERS (as an earlier commenter noted) as if they were margins or padding. The neat trick he didn’t mention: you can set BORDER-COLOR to TRANSPARENT so no need to match background colors or designs. 100% fills the remaining calculated width. You might have to keep an eye on your z-index layering when constructing to prevent mouse events from being captured in the wrong DIV.

     

    Copy & paste the code below to embed this comment.
  29. Whilst your CSS is clean and effective I thought I should point out a few difficulties that it exposes.
    Firefox has a well known and long established print bug exposed by the use of absolute positioning – content is curtailed, usually to a single page.
    Firefox mousewheel navigation (the autoscroll feature) is also broken.
    Neither of these may be considered show stoppers for some but developers might like to be aware of these issues.

    Copy & paste the code below to embed this comment.
  30. It looks like #main style is supposed to be the style for <div id=“right”>??

    Copy & paste the code below to embed this comment.
  31. Whilst the solution is neat and taught me things I didn’t know, I do have a worry about those users who can’t use a mouse for some reason.

    As exampled it seems impossible to get the DIVs to scroll down using keystrokes. It seems to me that a sighted user who cannot use a mouse would be unable to access the text at the bottom of both the DIVs.

    In a similar layout with frames tabbing would give focus to each frame in turn, and then the up and down arrow keys would scroll the individual frames.

    Does anyone have any suggestions?

    Copy & paste the code below to embed this comment.
  32. I really enjoyed the article.

    One issue I’m having in IE7 is getting a nested DIV to respect the calculated height of the parent div. In Firefox, setting the height property to 100% yields the desired result; that is, a div that consumes the entire height of the underlying element. In IE7, the height is ignored, presumably because the element’s height is calculated rather than specified in CSS.

    If someone has a workaround, I’d be grateful for tips.

    Copy & paste the code below to embed this comment.
  33. Great article rob. Very clear and concise, eays to follow. Good job.

    Copy & paste the code below to embed this comment.
  34. Thank you very much for this article. It solved me a month of headache.

    Copy & paste the code below to embed this comment.
  35. Love this technique. Had no idea this was possible. TY!!!

    Copy & paste the code below to embed this comment.
  36. The layout I’m stuck with has three fixed width columns. For over two years I have tried different ways to get IE6 (may it someday RIP) to extend the background of the right and left columns all the way to the bottom of the center column (where the varying height main content is). I don’t even have (or need or want) scroll bars.

    Maybe there’s an expression-free method to accomplish this somewhere out there, but I haven’t been able to find it. I’m just glad this approach works perfectly for my purpose.

    Thanks for taking the time to track down and explain a classic head-scratcher to this non-expert.

    Copy & paste the code below to embed this comment.
  37. Thanks, even if we’re seeing IE6 fade away, it is still very much a reality with Asian users.
    I have been looking for a cross-browser option of pinning a div to the viewport’s height, preferably without the use of jquery or equivalent.
    This is brilliant.

    Copy & paste the code below to embed this comment.