For some time I've been making websites, but have never really seen discussion of the proper usage of the container tags. I've seen any number of different types of content in the collection tags, but it usually seems that the page creator just picks a style they like and sticks with it.
The main discrepancy in my mind is that between
<p>
<div>
but I'd also like opinions regarding
<span>
and any others I may be forgetting.
I think page creators should use semantic markup, meaning that the markup they create should communicate meaning (and not presentation). <div> and <p> have different meanings. The former is used to define a division (or section) of an HTML page, the latter to define a paragraph of text.
I think, the meaning of the tags is something like this:
<p>Paragraph, usually just text</p>
<div>A block, containing anything</div>
<span>Just a simple non-blocking wrapper</span>
HTML was originally created to put the content of documents into some sort of structure. With that in mind, the <p> tag is supposed to hold anything that would be structured as a paragraph if the content of the page were to be turned into a printed document. The <div> and <span> elements are reserved as general use containers to facilitate formating and grouping of related elements to provide additional levels of structure, perhaps correlating to pages in a text document.
In some cases, <p> tags should contain other elements, such as anchor, <a>, image, <img>, and possibly lists, <ol>/<ul>, because they relate directly to the content of the rest of the paragraph and it makes sense to group them that way, or the text of the rest of the paragraph provides a more in-depth description. If there is not additional description of those elements, however, it does not make sense to place them in a paragraph simply as a convenient container; a div would be more appropriate. In general, a paragraph is supposed to contain one paragraph of text and any directly related or described elements. Nothing else makes much sense in a paragraph.
<p> is a block-level element that should contain a paragraph, comprised of text, inline elements that modify that text (<p>, <a>, <abbr>, etc.), and images.
<div> is a block-level element used to divide the page, almost always in conjunction with CSS styles.
<span>... well, I honestly don't use this tag that often. It's an inline element, and I use it usually when I'd like to apply styles to a portion of text that wouldn't benefit from using something with more meaning, like the <strong> and <em> tags.
I was tought to view <span> and <div> as the "tofu of webdeveloppement", since it has no real flavor but you can do virtually anything with it.
(X)HTML tags define what the text they're surrounding is. Is it and address, is it a link, is it a paragraph, and so on...
<div> and <span> are simply ways of getting to pieces of your site you normally can't get to. Like when you're trying to resize a | symbol. Fastest way I've ever found was to put a span around it, give it a class and then implement the CSS.
That's what they're good for, in my opinion. I'd be interested to hear more or even corrections on what I've written here.
The difference between these three (and many other) tags is their semantic meaning. The HTML standard includes both tags with specific semantic meanings (<p> for paragraphs, <em> for emphasized text, etc.) and tags without semantic meaning.
The latter are <div> and <span>, which are used to identify block- or inline-level content which needs to be identified (using, say a class= or id= attribute), but for which a semantically-specific tag does not exist. For example, one may write <p>Hi, my name is <span class="name">John Doe</span>.</p> — indicating that it's a paragraph (which the browser already has an idea how to handle) and that part of it's content is a name (which means absolutely nothing to the browser unless CSS or JavaScript uses it).
These tags are therefore incredibly useful both in adding additional information to an HTML document which doesn't fit within the semantic tags supplied by the standard (see the hCard specification for an excellent example) and for applying visual (CSS) or functional (JavaScript) structure to a document without altering its semantics.