Great overview/reference!

A tiny error:

em - relative to font size of the element

EM is relative to font size of the parent element. Not the element itself.

@equinusocio no, em is relative to the closest font-size definition.

html {
  font-size: 10px;
  border-width: 1em; /* 10px */
}
body {
  font-size: 20px;
  border-width: 1em; /* 20px */
}

It only follows the parent's if you are setting the element's own font-size

html {
  font-size: 10px;
  border-width: 1em; /* 10px */
}
body {
  font-size: 2em; /* 20px */
  border-width: 1em; /* 20px, not 10px, because it follows its own font-size */
}

https://www.w3.org/TR/css-values-3/#font-relative-lengths

the font-relative lengths refer to the font metrics of the element on which they are used. The exception is when they occur in the value of the font-size property itself, in which case they refer to the computed font metrics of the parent element

In realtà sono giusti entrambi i casi perche comunque se il font-size non è dichiarato nell'elemento stesso l'em fa riferimento alla prima dichiarazione che trova nei parent, fino ad arrivare alla root e al font-size impostato dal browser. Quindi di default, è relativo al font-size più vicino si, ma che per forza è un parent (se non espresso appunto nell'elemento stesso)

You probably wanna add ic into the Relative Units section which is relative to the width of the CJK "水" (WATER, U+6C34) character in the element's font.

The last link inside the gist also seems to be dead.

Thank you for this very nice overview!

About "Media Queries in em only" - that seems to be more of a myth than a fact or an actual recommendation.

Way back there were zoom problems in browsers, but in the 2012 article you cite the author has added this to the article in 2015:

You should note that the zooming behavior has long since been made consistent in browsers (i.e. fixed). Keep that in mind if you cite or otherwise put stock in this post. I never meant to be absolutely prescriptive about ems as units—use whatever feels natural and appropriate for your design strategy.

In other words: Media Queries in px are no problem.

Also there is a question about this from Stephen Hay to Ethan Marcotte right at the beginning of the "Speaker Panel | CSS Day 2014": https://vimeo.com/102847187#t=30s

Ethan prefers MQs in em but thinks that "it's more stylistic" and that "pixels are just fine".

Thanks for the post, and thanks @pmmueller for adding your note about this!

Why not use ch and ex frequently? They seem perfect for width and height

Great post.. Thanks

Excuse the dumb question but what do you mean by Print?

Excuse the dumb question but what do you mean by Print?

@marwanbayoumi You can use CSS to style a page differently whenever it is printed. In such cases, using a physical unit may be more appropriate. See these resources for more on this topic:

1. A Guide To The State Of Print Stylesheets In 2018 (Smashing Magazine)
2. Printing (MDN Web Docs)

Why not use ch and ex frequently? They seem perfect for width and height

@bn-l This guide is a bit bias toward web apps. When wanting to maintain fixed sizing and spacing throughout a user interface, it may be better to primarily use rem, as that provides a way to be flexible and consistent.

In the case of content sites, ch and ex units may occasionally be more appropriate. For example, if wanting to stay within the recommended 45–90 characters per line to improve readability, you could do this:

p {
 width: 45ch;
}

To calculate a similar width with another unit would involve math or trial-and-error. In this case, ch is a very straightforward solution.

@basham Practices about scaling the elements for high DPI screens like smartphone screens would be good. Really the use o fixed pixels becomes a mess when the site is displayed in a screen with different DPIs, which always happen when the website is public.

@basham Practices about scaling the elements for high DPI screens like smartphone screens would be good. Really the use o fixed pixels becomes a mess when the site is displayed in a screen with different DPIs, which always happen when the website is public.

@HMaker Do you have any recommendations in this area or have any suggestions for such a resource? I haven't studied this topic. Don't know if there's too much more than a resolution media query.

"Only use em within media query definitions, never pixels." - the article from 2012. I think this has been fixed or am I wrong?

Isn't it more responsive if we use rem for font-size rather than em.?

This document should be updated, it shows up quite high on Google but contains many outdated concepts, for example based on limited feature support, even when caniuse now shows very good support (see for example vmin/vmax or rem on "usare relative" view, which clearly shows unsupported browser versions are almost completely unused at this point).

The vmin/vmax section in particular could use some expansion on the use case, and the em section should probably recommend the use of rem, which doesn't yield the same issues as em, since it doesn't inherit from parent elements in cascade.

Furthermore, the use of percentages is promoted as though it would solve any problem, but that is not the case and it's important to redirect the user. See for example this question on SE.

@berzi there's also this article that suggests not to use relative units on layout. altho, I don't know how relevant it is.

That's what I've been looking for 😃

For media queries do we need to still use em vs rem ?

Verify Github on Galxe. gid:mGErGTSTvcAM2BuipcSWyn

nice/ but this wont work perfectly for media queries

vw/vh should be updated with lvh, lvw, dvw, dvh and the co.

Shouldn't we use 'rem' than 'em'?