box-sizing issue with input[type="search"]

[patrickhlauke]

-webkit-box-sizing: content-box;
box-sizing: content-box;

appears to be making search inputs wider than they should be - see twbs/bootstrap#17379

[BYODKM]

#358

[patrickhlauke]

So, maybe @necolas can confirm (as I just read through the linked issues there, with their very terse comments): is the fact that box-sizing: content-box results in search inputs being larger than their equivalent text inputs intentional then, and won't be fixed? /cc @cvrebert

[BYODKM]

Normalize.css is for normalize browsers' default style. Which means you can not always get what you want. You can have a small reset css for input[type=search] in your own project as usual. The same as the h1, fieldset and so on.

[patrickhlauke]

How about normalizing it to what should be the default, according to https://www.w3.org/Bugs/Public/show_bug.cgi?id=28784 ? I'm fine overriding it myself, but effectively here you're normalizing it against all current consistent (with exception of Firefox, but that's being fixed now too) default styling in browsers...

[FagnerMartinsBrack]

I am watching this repository for a long time, and as far as I know there was no discussion regarding search inputs being wider than they should be. Most discussions regarding that style is about it being harder to override, because one can't just include normalize and go for * { box-sizing: border-box; } and expect the search inputs to be overriden.

Other than that this seems pretty legit and worth looking into, as long as it doesn't cause a regression from the first bullet point.

[patrickhlauke]

As there seems to be little interest here in addressing this, closing. We fixed it back to browser defaults in Bootstrap, FWIW.

[necolas]

Sorry for being absent. I agree that this is an annoying quirk of normalize that should be fixed. It looks like the original reasoning doesn't stand up, and I'd rather be pragmatic about it anyway.

[patrickhlauke]

Sorry for interpreting silence for lack of interest/agreement. Wasn't sure how best to formulate a visual test, but gave it a go in the PR anyway.