>Another thing: It's Ok for some elements to be different between browsers.
This is one of those lessens that took me a while to learn. I spent a lot of my time early in my career trying to get every pixel the same between between IE and Firefox until someone asked my why I cared. I never gave the "why" much thought until that point; it was just something I felt I needed to do. As long as the site functions correctly and looks good in the various browsers, getting them exactly the same is nothing but a fool's errand.
To me it feels like the author of that project didn't get the point of resets.
The problem without resets is not that some client/manager/designer wants to see a pixel perfect page, but that differently sized default elements in different browsers can cause whole layouts to break / become non-functional.
If your layout is robust enough to compensate such variations, that's fine and you probably don't need a reset CSS at all, but the idea of reset CSS is not about efficiency in terms of lines of code, but about efficiency of developer resources. Developing something on top of a 100% defined state is much easier than constantly having to worry about different initial states.
Originally loads of people in industry came from print, so I can completely see how pixel perfect layouts were seen as desirable (hell, loads of our idiolect comes directly from print media, show me where a webpage folds over!) and conversely why they've died out over time as a generation of people who grew up with the internet have come through.
No kidding. I don't support IE at all, much less slave to make pages look the same in it, because I want to. I don't try to make things look the same in desktop browsers because I have some ideological devotion to the idea. I do that because otherwise, the clients whine.
Every time I see someone go, "Hey, have you ever thought of a page design, like, not being consistent between browsers?" and seeming to expect readers' minds to be blown, it makes me want to ask, "Have you ever considered doing this, like, as a job?"
Maybe I have just been lucky, but I have never had a job in which my only function was to translate a design doc or mockup into a website. I couldn't imagine working in that type of environment for long. Part of the job has always been to talk through design choices with bosses or clients because it is easy for a non-developer to make a decision they don't feel strongly about that would add a multiplier to the development timeline. So my original comment wasn't purely about my own preferences. It was also realizing the benefit of pixel perfect design is negligible and that I should be willing to argue against it when the issue is brought up.
I worked with a manager who had a degree in graphic design. He would flip out if the design was even half a pixel off. We came to call him Pixel Perfect Peter (his name wasn't Peter).
I'm a designer and frontend developer. I would find it odd for the end result of an HTML build to look any different from the design. I can't think of a reason why a developer (who's not involved in the UI or UX stages) would need to change anything, even by a pixel. That may also lead to problems down the road. If the mockup uses certain dimensions, paddings, line-heights, etc... why not translate them as indented? Especially in 2018. It wasn't fun making websites cross-browser friendly 20 years ago. Even 10 years ago.
Why isn't the developer involved in the design? Might help to point out elements of the design that would take large amount of effort to implement faithfully.
In fact why isn't the design done as a CSS to begin with?
> I can't think of a reason why a developer (who's not involved in the UI or UX stages) would need to change anything, even by a pixel.
I guess you supply your developers fully documented fluid-resizable, responsive, dynamic mockups with all possible states? That’s one lucky developer.
As a designer and developer, most (extremely talented!) designers that I end up working with still send me their designs as static PSD’s, and they also don’t understand enough HTML to know which design-choices are costly and which are not, and which don’t scale well with the screen / text / content, etc.
If the designer and developer don’t work together at the UI/UX stage then it’s completely understandable for the coded-up site to not be a pixel perfect carbon copy.
I generally supply mockups for different breakpoints from Sketch (via Zeplin). Moving away from making web UI in Photoshop was a game changer for myself, and the other developers. Dimensions/spacing is easy to read, and nearly all styles are translated to CSS. Highly recommend telling your designer friends to try software like Sketch or InVision Studio.
You mean the original author doesn't need more? I'm all for minimalism, but my minimalism doesn't necessary mean everyone's minimalism.
And while the readme mentions you might not need it, or you might need to add to it, IMHO if it's not something that meaningfully solves the problem for a majority of users, and if it's trivially small then it probably isn't that useful.
It's not really useful, no. But it's interesting. I read the readme and the five lines of code as a blog post. It makes a couple good points. First, there's no need to reset elements that aren't being used. Second, a reset file is redundant by nature since it's overwriting a "reset" provided by the browser, that's subsequently overwritten by any custom styling. But yes, in virtually every scenario it's more practical to not think about it and use normalize.
I got a completely different idea from reading the description. He's saying the whole point why this thing is small is that you don't need to keep stacking changes to the same properties, i.e. why do this:
He argues that you can just drop the first rule because you're going to be setting it to something else later anyways; you should only reset properties that are likely to be final and only tags you actually use.
Ideally, I don't disagree with the concept behind that approach. Clean, streamlined CSS is nice because it's generally easier to work with. Browser page inspectors help with that, though. You aren't searching up and down a file sheet to figure out what rules are applicable in a given context.
Resets (normalize.css or any other) are about consistency. Both between browsers and between projects. When you import one, you know what you're getting and don't have to think about it outside some edge cases. This micro reset eliminates that and forces you to figure out your reset needs each and every time. Odds are, you'll miss something at some point. That's especially likely when it comes to accessibility; good resets try to take it into account to avoid screwing users who rely on screen readers and other assistive tech, for instance. They don't always manage it, but their shortcomings are at least consistent. That makes it easier to fix.
Besides, there are already a lot of tools and options to accomplish the stated goals. You could wrap less commonly used reset rules in conditionals and set variables to disable certain ones (forms being the example given). Build tools have improved a lot in recent years. You can use things like cssnano[0] to help you optimize your stylesheets when they're built.
On the other hand, how important is this anyhow? Browsers are highly optimized. Having tons of duplicate media queries throughout a stylesheet (common when dealing with preprocessors and media query mixins) doesn't impact performance, for instance.[1] No reset is going to have a measurable performance impact in adding a handful of duplicate selectors. As for file size, gzip is extraordinarily efficient when it comes to dealing with property and selector names. If you're implementing something like this for performance reasons, the time spent crafting the smallest possible reset for each site you work on would be better spent elsewhere.
It goes both ways. One can argue that setting up css build tooling would be time better spent elsewhere (and if you ever did corporate or legacy work that can be painfully true)
In many cases, a mindset of not doing needless extraneous work is the result of having seen crushing complexity, and it's a mindset that goes a long way. As the saying goes, better to be the master of your tools than a slave to them.
That seems like an argument to post-process the CSS and avoid double-setting properties.
The point of a reset/normalise stylesheet is to create a coherent and consistent baseline for work, in theory you could actually stop there. Or just alter the reset itself.
That will force you to use completely different CSS layout model. Yes today is common practice the use of box-sizing: border-box and it should be the first line of code when you set your layout. I don't what to force anyone to use box-sizing: border-box.
"It sounds like you're saying today it's absolutely something you should use but you don't want to force it?"
Yes, correct.
Let me explain.. that is the most important decision you make when you plan your CSS layout. But is your decision to make. You can absolutely add this in the reset. Feel free to add stuff in the reset.
You don't need to reset things you won't use... then proceeds to reset tables, table properties, and iframe borders? This is the most nonsensical thing I've seen on here. Also resetting all h elements to weight normal when they're the most likely elements to have non-normal weights.
Persisting an npm package to your project for 5 css rules is a far more egregious case of "you might not need..." than a simple css reset could ever be.
at the start of every project and call it a day. That was my reset.
Later I found that was considered a bad practice, but honestly? I found CSS is very unlikely to be the bottleneck for performance in most things I deal with nowadays.
Anyway, I like this micro-reset but I rarely start css from scratch these days, otherwise I would just memorize these rules.
I have always been strongly opposed to CSS resets and global CSS styling rules in general. There really is bo need for it if you know how to isolate your classes correctly. Some Atlassian plugins create global styling rules and they tend to break things.
The problem isn't with "isolating classes correctly" (how do you isolate classes correctly in CSS' flat namespace?).
The problem is that even such basic things as lists and paragraphs will look differently in different browsers (different offsets, sizes, line heights etc.). Unless you cover every single element with a "correctly isolated CSS class", you need a global reset.
I recently started using CSS Grid with a view to writing concise, maintainable CSS that wasn't 'Jenga CSS' that people only add on to. I also wanted to get away from the 10000+ lines of cruft that many projects seem to think you need.
All was going well with the layout but I wanted to avoid margins and padding, instead I wanted to use grid gaps.
Initially I did not have the 'bravery' to strip out all 10000 lines of cruft, I assumed some of it was 'important'.
However, just 'for a laugh' I thought of getting rid of the normalize.css buried in there, complete with rules for things like IE6/7/8. Much to my surprise my layout suddenly looked awesome, all margin and padding issues resolved. I checked on a few browsers and all is looking very good.
The biggest bonus was on the input boxes. These now work much better for layout and usability on all screen sizes. I can use things like the 'size' attribute and make the boxes sensible for what is going to be entered.
Fluid width type and no fixed breakpoints are also a huge discovery along with CSS variables. I also gave up on divs and spans, not to mention class tags. Instead of these empty calories I use HTML5 tags with schema.org attributes. If a product has a name, description, price or other attribute, I mark it up for the screen readers first then the CSS can work on those attributes. No need for 37 layers of containing 'div' boxes each with class tags for some lame 'Foundation' style framework.
It seems absurd to remove the margin/padding and then have layers of Jenga CSS on top to get it to be as per the some designer's drawing.
The major benefit is in speed, I am sure that people who have been creating this unmaintainable CSS bloat for the last decade or find my approach heretical but the scores on the doors of what 'Lighthouse audit' says don't matter to them either.
I urge anyone to get rid of the resets as well as the inflatable arm bands and to let go of the hand-rails. Some people have never ventured out without their CSS resets, a bit like those children that have never gone anywhere without being in a car.
I like the reasoning behind this, but am surprised it doesn't include resets for the (fairly opinionated) user agent styles for lists, which I'd argue for most sites get used more often than tables.
There's no such thing, to my knowledge, as a "browser reset"; there is a standardized normative default user-agent style sheet for HTML4 as specified in CSS level 2.
I agree with everything else the author is saying, though. But there's no point in this existing when normalize.css does. Even then, you really shouldn't be using that in my opinion for the same reasons the author specifies.
This is one of those lessens that took me a while to learn. I spent a lot of my time early in my career trying to get every pixel the same between between IE and Firefox until someone asked my why I cared. I never gave the "why" much thought until that point; it was just something I felt I needed to do. As long as the site functions correctly and looks good in the various browsers, getting them exactly the same is nothing but a fool's errand.