collapsing margins
Truly understanding css is a rather long and difficult process. Also, getting something to work in css is different from fully understanding why it works. For me personally, the magic of the collapsing margin has always been one of the foggiest css areas around. So let's take a closer look at this murky little css behavior, and more importantly, why it causes damage to my hair from time to time.
It's not too big a deal if you don't understand their purpose, as you can work around them quite easily, but collapsing margins are there for a reason. The actual theory behind collapsing margins is a little too complicated to summarize, so for their precise behavior I'll just redirect you to the full w3 explanation of collapsing margins. In short though, I'll quote their definition of the general behavior of collapsing margins:
collapsing margins means that adjoining margins (no non-empty content, padding or border areas or clearance separate them) of two or more boxes (which may be next to one another or nested) combine to form a single margin.
Even shorter, when two consecutive elements have a vertical margin defined, those margins will overlap whenever they border on each other.
everything has a reason
p {margin:1em;} At first this might sound like strange behavior, but there is actually a good reason for collapsing margins, and I gladly made use of it in my quest for finding the differences between margins and paddings. In its most simple form, collapsing margins were conceived to control the spacing between paragraphs of text. If you consider the css above without collapsing margins, this rule would result in 2em spacing between each paragraph. Something you wouldn't exactly want or expect from a rule like the one above.
So to fix this problem, the vertical margins were made to collapse, forming a gap of 1em between each paragraph. Which is pretty neat actually. But they didn't stop there. A similar problem arose when paragraphs were dropped in a li element. Of course you'd want the first paragraph to line up with the top of the list element, so rather than keep collapsing margins restricted to siblings, they expanded the behavior to adjoining elements (fe, the parent and its first child). If you want some visuals with that, you can consult Meyer's article on the subject, which also gives a broader explanation of the problem.
breaking the collapse
Especially the second form of collapsing margins can be troublesome. Depending on the styling of the element, you'd want this type of collapsing to happen. The good people who wrote up the css rules realized this and added some rules to break the collapse. You could either use a padding or a border (so the margins wouldn't join anymore). Both cases are related to some sort of visual separation of the element. For example, if you give the list element a border, you'd want the elements inside to keep their distance from that border, increasing readability.
What they failed to fix is the case where the containing element is simply given a different background style. You still create the visual effect of separation, but your margins will fail to collapse, placing the first sibling against the top of the container, creating a cropped and ugly layout.
weighing the solutions
Breaking the collapse through paddings or borders is a bad idea. For one, you add at least 1 pixel to the separation of content and border, which might clash with the layout of elements above or below. And yeah, it's only 1 pixel, but I'd rather not have that difference. On top of that, I don't want to be adding paddings where they are not needed at all. And furthermore, applying a border means setting it to the same background color of the parent. If the color changes, two values have to be adapted, hurting flexibility of the css and introducing a bigger chance of errors.
One solution not described in the Meyer article is setting an overflow:hidden to the parent, but this only works if you don't need to place something outside the container. Also, I'm not too fond of adding an unneeded behavior to an element just to fix an unrelated problem. And again, it hurts the adaptivity of the css to changes further on in the development process.
This leaves us with no perfect solution, which is a real shame.
what we need
It would be helpful to check for different background settings between adjoining elements, but that sounds like an extremely fickle solution to me (though I am sure that would fix a whole lot of problems already). I'm not sure this will fix all occasions where collapsing margins are hurtful.
Better would be to give us, css people, control over collapsing margins. Give us a property to leave or break the collapse. That way this becomes our responsibility, rather than our problem.