CSS Specificity - Defining Classes & IDs

问题

I've been reading up on specificity and, quite frankly, I'm surprised that I didn't know about this properly before, since I've witnessed the very issues that specificity can give if CSS is declared in the wrong manner.

So, I've been doing a little research on the subject, and I now understand the rule for calculating specificity. However, during my findings, I've been left with three questions which I was hoping you guys could help me out with.

Firstly, when relating to CSS specificity, I've noticed that one source includes pseudo elements in the calculation, while another tells you to leave them out of the calculation. Is this purely an optional thing, or should I include them?
Secondly, is there any specific reason why I should declare classes above identifiers? For instance, I've already sectioned my development CSS file, and in the 'footer' section, I have #footer, #footer-container, .grid and .one-third, declared in this order. Is this perfectly fine, or should I switch them around? The defined rules are below, to show that both classes don't contain any conflicting properties:
```
#footer {
   background: #e4e4e4;
   border-top: 1px solid #eeeeee;
   border-bottom: 1px solid #666666;
   overflow: hidden;
   padding-bottom: 1em;
   width: 100%;
}
#footer-container {
   margin: 0 auto;
   width: 100%;
   min-width: 1000px;
}
.grid {
   margin-right: 2.1%;
   float: left;
   display: inline;
   position: relative;
}
.one-third {
   width: 31.9%;
}
```
Lastly, just a quick question about listing CSS properties in alphabetical order. I fully understand that this is optional, but using my declaration for #footer above as an example, is there any preference for border, margin or padding, as in alphabetical (bottom; left; right; top), or declared as written in shorthand (top; right; bottom; left)?

Thank you very much!

回答1:

Firstly, when relating to CSS specificity, I've noticed that one source includes pseudo elements in the calculation, while another tells you to leave them out of the calculation. Is this purely an optional thing, or should I include them?

This is news to me; it looks like pre-CSS2.1 pseudo-elements were originally supposed to be ignored when calculating selector specificity. The article that says to leave them out of the calculation was published almost a decade ago, right in the middle of CSS level 2's refactoring. Here's what the 1998 CSS2 recommendation says (which, for reference, is also quoted in that article, but the URL that it links to redirects to the latest revision of the spec where it's changed):
A selector's specificity is calculated as follows:
- count the number of ID attributes in the selector (= a)
- count the number of other attributes and pseudo-classes in the selector (= b)
- count the number of element names in the selector (= c)
- ignore pseudo-elements.
In CSS2.1, as well as the latest Selectors standard, pseudo-elements must be counted like normal type selectors when calculating specificity.

Perhaps what's even more curious is that CSS1 had the same rule for pseudo-elements in terms of specificity as the current standard:

Pseudo-elements and pseudo-classes are counted as normal elements and classes, respectively.

This leads me to believe that the change in CSS2.0 was ultimately reversed in CSS2.1 so it wouldn't break existing (legacy) stylesheets that happened to rely on the behavior, and/or because it simply didn't make sense not to include pseudo-elements since you apply styles to them like you do with actual elements.
Secondly, is there any specific reason why I should declare classes above identifiers? For instance, I've already sectioned my development CSS file, and in the 'footer' section, I have #footer, #footer-container, .grid and .one-third, declared in this order. Is this perfectly fine, or should I switch them around?

There should be no difference whether you place your classes before your IDs, as long as your two classes follow the order they are in and your two IDs, likewise. Even assuming that all of these rules happen to match the same element, the IDs should take precedence over the classes for any properties that need to be overridden.
Lastly, just a quick question about listing CSS properties in alphabetical order. I fully understand that this is optional, but using my declaration for #footer above as an example, is there any preference for border, margin or padding, as in alphabetical (bottom; left; right; top), or declared as written in shorthand (top; right; bottom; left)?

If you are declaring the longhand properties separately, then it's entirely up to you how you want to order them. I would certainly follow the order already laid out by the shorthands for the sake of simplicity.

FYI, the order in the shorthands is laid out the way it is because if you literally connect the dots, the order is actually clockwise starting from the top.

回答2:

Many would argue not to use IDs at all for styling to avoid issues with specificity. You CAN use classes over and over or you can just use them for one element like an ID. I agree that it is an unnecessary issue to battle with when its just as easy to use a class. And before you say "but sometimes I cant change the markup." let me just say attribute selector. These may be a little more to type but to avoid problems with specificity I think its worth it. So

div[id='yourID'] {

}

and you dont have to worry about not being able to overwrite this later. This is especially helpful when you work on a team.

来源：https://stackoverflow.com/questions/21455250/css-specificity-defining-classes-ids

标签

css

css3

css-selectors

css-specificity