Why does this happen?
Because mobile devices usually need to change the screen dimensions when a keyboard is shown, they come with some restrictions on when a focus event can be triggered. This is intentionally limited browser behavior on touch devices to guarantee a good UX and avoid performance issues of resizing the browser window ( = potentially very expensive style recalculation).
There's some cross browser differences in how far these restrictions go, and some browsers have bugs on top of that.
But in general they all require an actual user interaction to have happened not too long before the focus is programmatically triggered. Using btn.click()
is not the same as an actual touch event, and so the browser will ignore it seeing there was no recent touch event.
On a tracking issue on this behavior for Webkit, Apple provides a motivation:
We (Apple) like the current behavior and do not want programmatic focus to bring up the keyboard when you do not have a hardware keyboard attached and the programmatic focus was not invoked in response to a user gesture. Why you may ask...because auto bringing up the software keyboard can be seen as annoying and a distraction to a user (not for your customers, but for everyone not using your app) given that:
- We bring up the keyboard, which takes up valuable real estate on screen.
- When we intent to bring up the software keyboard we zoom and scroll the page to give a pleasing input experience (or at least we hope it is pleasing; file bugs if not).
This similar issue with the autoplay
attribute, which also requires a gesture (or page load) to have happened recently enough, could be helpful in understanding the kind of problem.
How to work around these limitations?
Most of the time
Likely you don't need to, as most code that would trigger focus is running as a response to a user action and will be allowed to trigger focus. I didn't find data yet on how big the window of time is, but I guess it's big enough that in regular cases it's not a problem (unless you set timeouts of course).
A possible problem for those cases is the script would run so long that the event is already considered too long ago. In fact that's what happened in the related autoplay issue, where the code would request some network resource, and only after the response would trigger the video to play. Depending on the speed of the network/device the video would sometimes auto play, and sometimes not.
Technically the same could happen with an input that is only focused after a network delay making it not show the keyboard. As long as your code doesn't do any network requests you won't have this kind of problem.
On page load
This is definitely not a "fix" for the problem, but you can do a best effort to manage focus on page load by using the attribute specifically made for it. This still won't make a keyboard appear when it otherwise wouldn't.
The autofocus
attribute is at least partially supported on all browsers.
Perhaps just calling focus()
directly in a script on page load works, but again there's a chance that this code runs too long after the page started being displayed, especially on slower devices. This probably also happens when adding HTML with the autofocus
attribute programmatically (e.g. React). If the initial HTML contains the autofocus
attribute on the right element that shouldn't occur.