UPDATE: The issue seems to stem from having many select elements on a page. How random is that?
So here's the issue. On iOS 7 Safari, when tapping the a text input on my site, the keyboard opens then freezes the OS for about 2-5 seconds then finally scrolls to the input. After this happens once, it never happens again until you refresh the page. I've looked all over the place, and yes, iOS 7 Safari is super buggy, but lets try and see if we can figure this out.
Note: This does not happen in any other mobile browser or any previous iOS Safari. It happens both on the ios 7 iphone and ios 7 ipad.
I will list everything my friend and I have tried so far:
Removed the ability to add event handlers in jQuery. (Note: all our event handlers are assigned through jQuery except for unload and onpageshow).
Removed the jQuery autocomplete script from the inputs.
Removed all JavaScript from the inputs.
Removed all third-party libraries being added on the page by rejecting the domains on the Mac.
Switched back to previous jQuery versions. The last one we could actually use before nothing worked was 1.7.0.
Switched back to previous jQuery UI versions.
Changed input event handling to delegate and live, instead of on('click')
Removed all CSS classes.
Removed all CSS from the page. Note: The response time for the OS this went down to 1-2 seconds but still happened.
Does anyone have any ideas?
Thanks a bunch!
(There are some somewhat-effective solutions, see near the end of the list)
At my company we are also suffering from this. We filed an issue with Apple but have heard mum.
Here are some interesting jsfiddles to help illustrate some of the issues, it definitely seems to revolve around the number of hidden fields, and textareas do not seem to be affected.
From debugging efforts, my guess is that there is some functionality trying to detect if an input is a credit card or phone number or some special kind which seems to cause the locking behavior. This is just one hypothesis though..
Summary:
On a page with a form containing named input elements inside containers that are marked "display: none", the first press on an input in that form has a very noticeable delay (20sec-2min) between the keyboard coming up and the input being focused. This prevents users from using our web app due to the enormous time spent with the ui frozen waiting for the keyboard to respond. We have debugged it in various scenarios to try and discern what is going on, and it appears to be from a change in how iOS7 parses the DOM versus how it did on iOS6, which has none of these issues.
From debugging within Safari's Inspector with the iPad connected, we found that iOS7 provides much more information about the (program)'s activities, to the point that we found that _CollectFormMetaData is the parent of the problem. Searching for meta data causes massive churn that increases more than linearly along with the number of hidden containers containing inputs. We found that _isVisible and _isRenderedFormElement are called far more than they reasonably should be. Additionally, if it helps, we found some detection functions relating to credit cards and address books were large time consumers.
Here are some jsFiddles for illustration. Please view them in Safari on an iPad running iOS6 and then on an iPad running iOS7:
http://jsfiddle.net/gUDvL/20/ - Runs fine on both
http://jsfiddle.net/gUDvL/21/ - Just noticeable delay on iOS 7
http://jsfiddle.net/gUDvL/22/ - More noticeable delay on iOS 7
http://jsfiddle.net/gUDvL/29/ - VERY noticeable delay on iOS 7
http://jsfiddle.net/gUDvL/30/ - Same as 29 but with none hidden - no delay on iOS 7
http://jsfiddle.net/gUDvL/38/ - Same as 29 but further exacerbated
http://jsfiddle.net/gUDvL/39/ - 99 hidden inputs, one visible, one separately visible
http://jsfiddle.net/gUDvL/40/ - 99 hidden textareas, one visible, one separately visible
http://jsfiddle.net/gUDvL/41/ - 99 hidden inputs, one visible, one separately visible, all
with the autocomplete="off" attribute
http://jsfiddle.net/gUDvL/42/ - 99 hidden inputs, one visible, one separately visible. Hidden by position absolute and left instead of display.
http://jsfiddle.net/gUDvL/63/ - Same as gUDvL/43/ but with autocomplete, autocorrect, autocapitalize, and spellcheck off
http://jsfiddle.net/gUDvL/65/ - Same as gUDvL/63/ but with cleaned up indentation (seems slower on iPad)
http://jsfiddle.net/gUDvL/66/ - Same as gUDvL/65/ but with display none via css again instead of DOMReady jQuery
http://jsfiddle.net/gUDvL/67/ - Same as gUDvL/66/ but with TedGrav's focus/blur technique
http://jsfiddle.net/gUDvL/68/ - Same as gUDvL/66/ but with css driven text-indent instead of display:block again (noticeable improvement - reduction to 2-3 secs for initial focus)
http://jsfiddle.net/gUDvL/69/ - Same as gUDvL/68/ but with TedGrav's focus/blur re-added
http://jsfiddle.net/gUDvL/71/ - Same as gUDvL/66/ but with js adding a legend tag before each input. (noticeable improvement - reduction to 2-3 secs for initial focus)
<input type="text" autocomplete="off" /> (links to jsfiddle.net must be accompanied by code..)
(We should note that having the iPad connected to a Mac with Safari's debugger engaged dramatically emphasizes the delays.)
Steps to Reproduce:
Load any of the above jsfiddles on the iPad
Press an input to gain focus
Watch screen until you can type
Expected Results:
Expect to be able to type as soon as the keyboard pops up
Actual Results:
Watch the keyboard pop up and the screen freeze, unable to scroll or interact with Safari for a duration. After the duration, focus is given as expected. From then on no further freezes are experienced when focusing on inputs.
tl;dr technique summary
So overall there are a couple proposed fixes from various answers:
Don't hide the divs with display: none - use something like text-indent
Short circuit Apple's metadata scanning logic - many form tags or legend tags seem to do the trick
Auto focus/blur - Did not work for me but two people reported it did
Related threads at Apple:
https://discussions.apple.com/thread/5468360
There seems to be a problem with how IOS handles the touch-event for inputs and textareas. The delay gets larger when the DOM gets larger. There is however not a problem with the focus event!
To work around this problem you can override the touchend event and set focus to the input/textarea.
document.addEventListener("touchend", function (e) {
if (e.target.nodeName.toString().toUpperCase() == 'INPUT' || e.target.nodeName.toString().toUpperCase() == 'TEXTAREA') {
e.preventDefault();
e.target.focus();
}
});
This will however create a new problem. It will let you scroll the page while touching the input/textarea, but when you let go, the site will scroll back to the original position.
To fix this, you just need to check if any scrolling has occured, and surround the preventDefault and target.focus with an if statement.
To set the original position, you can use the touchstart event.
document.addEventListener("touchstart", function (e) {
... //store the scrollTop or offsetHeight position and compare it in touchend event.
}
EDIT Me and a colleague have improved it a little bit, and it works like a charm.
var scroll = 0;
document.addEventListener("touchstart", function (e) {
scroll = document.body.scrollTop;
});
document.addEventListener("touchend", function (e) {
if (scroll == document.body.scrollTop) {
var node = e.target.nodeName.toString().toUpperCase();
if (node == 'INPUT' || node == 'TEXTAREA' || node == 'SELECT') {
e.preventDefault();
e.target.focus();
if(node != 'SELECT') {
var textLength = e.target.value.length;
e.target.setSelectionRange(textLength, textLength);
}
}
}
});
Struggled with this issue as well within an ios fullscreen which was inserting /removing pages containing a single input element. Was experiencing delays up to 30 seconds with only a single visible text input element on the page (and within the entire DOM). Other dynamically inserted pages with single or multiple text inputs in the same webapp were not experiencing the input delay. Like others have mentioned, after the initial delay, the input field would behave normally on subsequent focus events (even if the dynamic page containing the input element was removed from the DOM, then dynamically re-rendered/inserted back into the DOM).
On a hunch based on the above behaviour, tried the following on page load:
$("#problem-input").focus();
$("#problem-input").blur();
While the above executes immediately with no delay, the end result is no subsequent delays when the input gets focus via user interaction. Can't explain the reason behind this working, but it appears to work consistently for my app while other suggested fixes have failed.
I have the same freezeing problem.
I am not sure we're in the same situation.
here is my demo:http://tedzhou.github.io/demo/ios7sucks.html
In my page, i use a <p> element with onclick attribute as a button.
When user click on the button, page change to a textarea.
Then a click on it will freezes the browser.
The time freezing spent relevent to the numbers of the dom elements.
In my pages, there are 10000 elements, which make it freeze by 10+ seconds.
We can solve the problem by switching the <p> element to the real <button>, or reducing the nums of dom elements.
ps: sorry for my poor english. LOL
The main issue for me was with hidden fields. Made the form hang for 10-15 seconds.
I managed to get around by positioning the hidden form fields off the screen.
To hide:
position: absolute;
left: -9999px;
To show:
position: relative;
left: 0;
Met the same problem in quite complex application having many inputs.
Attached debugger to Safari iOS7 via USB and logged UI events. I see "touchend" event coming as soon as I am clicking on textarea (or any input) and in 10-20 seconds after that I see "click" being dispatched.
Clearly it is a bug in Safary as on other devices like Android or iOS6 there is no problem with the very same application.
It happens not only in iOS but in safari 7 for MAC OS (Maverics) too, I have found that the problem happens when you use a lot of div tags to contain inputs (or selects) within a form:
<div> <select>...</select> </div>
<div> <select>...</select> </div>
...
I changed the layout of my selects to use ul/li and fieldsets instead of divs and the freezze time was reduced drastically.
<ul>
<li><select>...</select></div>
<li><select>...</select></div>
</ul>
Here are two examples in jsfiddle:
freezze for 5 seconds
http://jsfiddle.net/k3j5v/5/
freeze for 1 second
http://jsfiddle.net/k3j5v/6/
I hope it might help someone
For me, this issue was being caused by user inputs being hidden on the page with display:none.
The workaround I used: instead of hiding inputs with display:none, I used jQuery's detach() method on document ready to 'hide' all the user inputs that were not being used. Then append() the inputs when they were needed.
That way no inputs had display:none on when the page was first loaded and so no delay occurred on the initial user interaction.
We had the same or a similar problem at my company. Whenever we displayed a large number of drop down lists and then a user clicked on a drop down, IOS 7 would freeze the page for a minute or two. After it unfroze, everything would work properly from that point forward.
This affected all input types. The large number of drop downs were actually hidden on first load - the user would initiate the display of the drop downs. Until the drop downs were displayed - everything would work fine. As soon as they were displayed, the next input click, even an input that had been working properly, now would cause the browser to freeze.
As others have noted, it seems that IOS 7 has a problem when parsing the visible inputs in the DOM after the user first interacts with an input. When the number and/or complexity of the elements/options/DOM are higher, the freeze is more pronounced.
Because it always froze on the initial user interaction, we decided to initiate a hidden user action as soon as we displayed the list of drop downs. We created a transparent button (it could not be hidden - it had to be "displayed") and initiated a click on it as soon as the user opened the drop down list. We thought that this would make IOS start parsing the DOM quicker, but found that it actually fixed the problem completely.
I have encountered this problem as well since I noticed many people are still having a problem with this I thought I'd put my solution.
Basically my solution is server side hiding of elements.
My page is ASP.NET so I wrapped my divs with the inputs with Panels and set these panels as Visible false.
This way if I click on an input the safari can't see all the other controls since they are hidden server side.
Of course if you want to make this work a little like clientside jquery you'll need automatic postback and an updatepanel somewhere.
This solution requires an effort but still its better than actually trying to fix a safari bug.
Hope this helps.
My answer might be slightly off the main topic, but I did arrive here after some searching as the scenario "feels" similar.
Issue:
My issue felt like a lockup in iOS, but not quite, since other elements on the page were still interactive. I had an <input type="search" /> element, that would not focus when I clicked into the field. But it would eventually catch focus after about 4-5 taps on the screen.
Additional Info:
My project is a hybrid app: WebView inside of an iOS app. The site is built with Twitter Bootstrap.
Solution:
I happened to also have the autofocus attribute set on the element. I tried removing that and it worked... no more consecutive taps to get the field to focus.
iOS 12.1.1 - December 2018
Here is a simple fix that worked in my case:
window.scrollTo(0,0) // attached to 'blur' event for the input fields
While it may not be ideal in terms of UX (especially if you have a form with many fields), it's definitely better than having 10+ second freezing time.
have you tried to turn off "Password & Autofill" > "Credit Cards" into Safari settings ?
After this operation it works fine. This isn't a final solution but maybe the problem's reason on iOS.
I'm trying to use TinyMCE in a widget but it fails. I think the problem is that view is still hidden when "viewAttached" is fired. It seems that TinyMCE has a bug/feature (read last paragraph) and can't be displayed when the target (textarea) is hidden (or inside a hidden div).
I got it working by doing the job in a setTimeout but it's crappy.
Is there a callback that I could attached to which is fired after the view is unhided (after the transition is completed)?
I found one solution:
Explicitly subscribe to the "isNavigating" observable of the router and add TinyMCE when "isNavigating" value becomes false.
Still : this has the effect of flickering - you see the textarea and then it is replaced by TinyMCE... but this is not a Durandal problem IMO.
Edit 1
Finally, I think the the best solution (for now... follow the link below for the thread on the subject) is to do a setTimeout(xyz(), 0) - I have seen a lot of people using this technique and it prevents the flickering.
https://groups.google.com/forum/?fromgroups#!topic/durandaljs/5NpSwMBnrew
Durandal does have a callbacks when you're using composition - you just put a function on to your viewModel with the correct name. In your case, you would use viewAttached:
Here's the docs:
http://durandaljs.com/documentation/Interacting-with-the-DOM/
I'm trying to use a custom video player NPAPI plugin (view FireBreath) inside an tabbed ExtJS application. The plugin lives in one tab, and the others contain presentations of other non-video data.
When switching from tab to tab, the element that contains the plugin is destroyed, and all plugin state is lost. Is there any way to configure an ExtJS tabbed panel so that the html contained in it is not altered when switching to another tab (just hidden)? The alternative is to re-populate the plugin state when returning to the tab, but this would be associated with an unacceptable delay (mostly while waiting for video key frames).
Thanks,
O
I don't know about your ExtJS approach, if you can solve it on that side that would of course be preferrable.
However, if you can't, you can avoid the reinitialization by moving the stream handling to a helper application that is running in the background. The plugin would launch it as needed and receive the stream data from it after registering for it.
The helper would be told when to kill a stream and possibly kill it by itself after some timeout (to avoid session leaks in case of crashing plugins etc.).
I was about to consider a helper application as recommended above, or look into rewriting the plugin to be windowless. Both might be more robust solutions for other JS frameworks.
Fortunately, the solution ended up being simpler than this, at least for ExtJS. By default, ExtJS sets "display: none" on the tabbed view's div whenever it is undisplayed, which calls the plugin destructor. After doing a little more looking through their enormous API, ExtJS has a parameter hideMode as part of the Ext.panel.Panel base class:
'display' : The Component will be hidden using the display: none style.
'visibility' : The Component will be hidden using the visibility: hidden style.
'offsets' : The Component will be hidden by absolutely positioning it out of the visible area of the document. This is useful when a hidden Component must maintain measurable dimensions. Hiding using display results in a Component having zero dimensions.
Defaults to: "display"
Setting the parent Panel that contains the plugin to hideMode: 'offsets' fixed the problem perfectly.
As we are facing GWT performance issues in a mobile app I peeked into Google Wave code since it is developed with GWT.
I thought that all the buttons there are widgets but if you look into generated HTML with firebug you see no onclick attribute set on clickable divs. I wonder how they achieve it having an element that issues click or mousedown events and seemingly neither being a widget nor injected with onclick attribute.
Being able to create such components would surely take me one step further to optimizing performance.
Thanks.
ps: wasnt google going to open source client code too. Have not been able to find it.
You don't have to put an onclick attribute on the HTML to make it have an onclick handler. This is a very simple example:
<div id="mydiv">Regular old div</div>
Then in script:
document.getElementById('mydiv').onclick = function() {
alert('hello!');
}
They wouldn't set the onclick property directly, it would have been set in the GWT code or via another Javascript library.
The GWT documentation shows how to create handlers within a GWT Java app:
public void anonClickHandlerExample() {
Button b = new Button("Click Me");
b.addClickHandler(new ClickHandler() {
public void onClick(ClickEvent event) {
// handle the click event
}
});
}
This will generate an HTML element and bind a click handler to it. However, in practice this has the same result as using document.getElementById('element').onclick() on an existing element in your page.
You can hook functions to the onclick event using JavaScript. Here's an example using jQuery:
$(document).ready(function(){
$("#div-id").click(function(){
/* Do something */
});
});
If you're interested in optimizing performance around this, you may need to investigate event delegation, depending on your situation.
A click event is generated for every DOM element within the Body. The event travels from the Body down to the element clicked (unless you are using Internet Explorer), hits the element clicked, and then bubbles back up. The event can be captured either through DOM element attributes, event handlers in the javascript, or attributes at any of the parent levels (the bubbling or capturing event triggers this).
I'd imagine they've just set it in a .js file.
Easily done with say jQuery with $(document).ready() for example.
Clicking on an element which has a Javascript handler makes the element go have a 'grey overlay'. This is normally fine but I'm using event delegation to handle the touchdown events of many child elements. Because of the delegation the 'grey overlay' is appearing over the parent element and looks bad and confusing.
I could attach event handlers to the individual elements to avoid the problem but this would be computationally very wasteful. I'd rather have some webkit css property that I can override to turn it off. I already have visual feedback in my app so the 'grey overlay' is not needed.
Any ideas?
-webkit-tap-highlight-color
To disable tap highlighting, set the
alpha value to 0 (invisible)
$('body').bind('touchstart', function(e){
e.preventDefault()
})
For those reading this question, another CSS property which will not only remove the overlay but also prevent the touch device from interacting with the user to assess a tap intention (which delays the timing of the event trigger significantly), is to use the cursor property.
a.notTappable {
cursor:default
}
Also, in the case the asker described where you have descendants of a parent and you only want the descendants to be tappable and you wish to IGNORE ALL MOUSE EVENTS OF THE PARENT, the following solution will you give you the best performance (by completely disabling the bubbling up to the parent). Read the webkit spec here.
.parent {
pointer-events:none;
}
.parent .descendant {
pointer-events:default;
}
I only mention this because the performance difference here is significant.