turbolinks/turbolinks-classic

Classic version of Turbolinks. Now deprecated in favor of Turbolinks 5.

turbolinks-classic's Issues

Set Accept-Encoding header to request gzip data

HTML compresses well, dont waste the users bandwidth!

Post request instead of get when hitting refresh

After making a post action which don't redirects you (error messages i.e), if you navigate to any link and then hit refresh, the browser (Chrome at least) tries to refresh the current url with a post request.

Thanks for all the awesome stuff you are sharing <3
Matt

Trigerring page:fetch event when fetching from cache

Shouldn't the page:fetch event be triggered when the page is fetched from the cache? Right now, the event only seems to trigger when the page is fetched from the server.

Incompatible with Chrome for iOS

There are some pushState-related differences from Mobile Safari in Chrome for iOS that cause Turbolinks not to work. Here's what I've found:

1. rememberInitialPage runs on the first call to visit in Mobile Safari, but in Chrome for iOS it runs as soon as the page loads.
2. In Mobile Safari, the window.history.state access in rememberCurrentState reflects the state that was set in the previous call to replaceState in rememberCurrentUrl. In Chrome for iOS, window.history.state is null in rememberCurrentState, so currentState is null too.
3. Because currentState is null in Chrome for iOS on the first call to visit, cacheCurrentPage fails. If you explicitly call Turbolinks.visit, the end result is the page never loads. If you click a link, it will fall back to a full page load.

is not working with masonry

I have using https://github.com/desandro/masonry in my project rails.

I have added this marvelous gem is incredible fast 👍 but masonry elements are not loaded successfully :(.

how we can solve this problem? Thank you very much!

Twitter Button breaks Turbolinks?

Hi,

I tried to get Turbolinks working on a project I'm playing around with but I couldn't figure why it didn't worked. Then I grabed this example project from Railscasts and searched along. It seems that as soon as I add an embeded Twitter button Turbolinks stops working. Is there any workaround or fix?

Thanks

302 redirects do not seem to work

It seems that when you click on a link where a 302 redirect is made, the content is changed but the url in the browser reflects the url before the redirect.

Example :

• When I click on http://example.com/carts/42/line_items?item_id=123
• I should be redirect to http://example.com/my_cart
• The content is the good one, but the url is http://example.com/carts/42/line_items?item_id=123 in the navbar

versions not tagged in repo

Possibility to replace meta tags

Hi,

Currently turbolinks replaces only body and title as far as I know.

I needed to also replace other meta tags such as opengraph and HTML5 microdata.

<meta property="og:title" content="Article title"/>
<meta property="og:type" content="article"/>
<meta property="og:url" content="http://news.com/items/1228"/>
<meta property="og:site_name" content="News Agency"/>
<meta property="og:description" content="news"/>
<meta property="article:tag" content="Science"/>
<meta property="og:image" content="https://url.to/thumb.jpg"/>

What would be the best approach? maybe use javascript?

suggestion for turboforms

I've got an implementation of turboforms working which hijacks all 'normal' forms on the page and sends then asynchronously, replacing the body with the result. This example is rails-specific but could be made generic:

extractUrlTitleAndBody = (content) ->
  doc = createDocument content
  url = doc.querySelector('meta[name=currentPath]').getAttribute('content')
  title = doc.querySelector 'title'
  [ url, title?.textContent, doc.body ]

setContent = (url, title, body) ->
  cacheCurrentPage()
  reflectNewUrl url
  changePage title, body

turboforms = () ->
  $('form:not([data-remote]),a[data-turboform]').each (index, el) =>
    el = $(el)
    el.attr('data-remote', 'true')
    el.bind 'ajax:beforeSend', (event, data, status, xhr) =>
      triggerEvent 'page:fetch'
    el.bind '[RemoteForm] ajax:success', (event, data, status, xhr) =>
      setContent extractUrlTitleAndBody(xhr.responseText)...
      triggerEvent 'page:load'

It raises two questions

1. is turboforms something people want?
2. this could be made as an extension to turbolinks (this is kind of how I'm hooking into it). The turbolinks api doesn't really allow for this at present but could easily be exposed to allow for extensions like this

Possible to disable for a whole page?

Perhaps a data-no-turbolink on the body element?

Link extension Filtering issue with digested images

If linked to digested image files (like in a photo gallery), the filter for .html extension does not work because it ends in something like ?dafdfeadfe , need to change match string to look before ?

Doesn't appear to respect data-confirm...

I have the following in my layout:

%a{href: user_applications_path, data: {confirm: 'Are you sure you want to navigate away from this page? All unsaved data will be lost.' }}

When this link is clicked, an alert is displayed, as-expected. However, if I click 'Cancel', the link is followed. The expected behavior is not to follow the link, right?

As a note, this was working with rack-pjax.

skip turbolinks for links with href="#"

Back button and reload can cause the wrong page to display

Misbehavior

1. Open a page that uses turbolinks. Call this Page 1.
2. Click a link to another page on the site using turbolinks. Call this Page 2.
3. Click the Back button to get back to Page 1.
4. Click the Reload button.
5. Click the link to Page 2 again.
6. Click the Reload button.
7. Click the Back button.

You will now be looking at Page 2 but Page 1's address is in the address bar.

This happens for me in Firefox 16, Chrome 22, and Safari 6.

Cause

Turbolinks' page cache ids (state.position values) start at window.history.length - 1. When you hit the back button, window.history.length doesn't change. Hitting reload makes Turbolinks restart its id numbering at window.history.length - 1. Certain dances involving back and reload can cause two different pages to have the same state.position in the browser's records, making Turbolinks pull the wrong page from its own cache.

In the case above, assuming you started with a new tab, window.history.length == 3 from step 2 onward. In step 5, Turbolinks tells the browser Page 1's state.position is 2. In step 7, Turbolinks tells the browser Page 2's state.position is also 2. Turbolinks puts Page 2 in its cache, then processes the popstate event to load Page 1. Since Page 1's state.position is 2, it just pulls Page 2 out of the cache, thinking it's Page 1.

Suggested Fix

Instead of window.history.length-1, you could use new Date().getTime()

Some inline script tags not being evaluated

If you have a script tag at the bottom of a form...to say initialize custom elements in the form. This script tag is not being evaluated from a turbolink. It is not a dom ready issue, it simply does not get evaluated by the dom writer.

Not sure what the issue is here, difficult to track down. If i put a script tag at the top of the body element it seems to evaluate it. its down further where it does not.

Issues with bootstrap plugins

Hi, I looking for the best way to do that.. don't know if it could be fixed in turbolinks, in bootstrap, or anywhere...

But, I recently created a question in stackoverflow, and found that's the strange behavior is because the way of modal plugins of bootstrap were implemented.

Maybe it is a possible hack to do with that, to made it work...

Safari 5 not supported

It appears Safari 5.1.7 does not support window.history.state which was introduced in commit 185dec1. Since it degrades gracefully I can understand if you don't want to support this version, but I wanted others to be aware of it in case they try to test Turbolinks behavior on this version.

Perhaps it should state somewhere in the readme which browser versions are supported? This can help when developers are testing and debugging on various browsers. If developers are unknowingly using an unsupported browser they would not notice JavaScript that is incompatible with Turbolinks.

Canceling confirm popup still goes to path in link

<a href="/users" class="btn btn-danger float_right" data-confirm="Are you sure?" data-method="delete" rel="nofollow">Cancel my account</a>

When I click on this link and then click cancel on the confirmation popup a GET request is made to /users instead of canceling the request.

If I click ok in the confirmation popup the delete is processed correctly. Using a full url instead of the relative path for the href, removing turbolinks, or adding data-no-turbolink fix this issue.

Based on the documentation shouldn't turbolinks catch this since I have data-confirm and data-method in the link?

This is the devise cancel my account link from the template it generates.

ignoring samePageLink means page is fully reloaded

If you're on / and then for example click the logo which is a link to / then we end up with a full page load because turbolinks is ignoring it.

Is this expected behaviour?

redirect_to in before_filter does not update address in address bar...

In my application (as in most applications), my ApplicationController handles authorization.

I have a before_filter method in my UserSessions controller:

before_filter :require_guest, only: [:new, :create]

which calls the method within ApplicationController:

def require_guest
  redirect_to user_path, notice: 'You are already registered and logged-in.' if current_user
end

This keeps people from trying to log-in when they're already logged-in.

When turbolinks is active, the redirect happens in the browser window and an alert is shown to the user (expected behavior). However, the URL in the address bar still shows 'site.com/login' (the URL for UserSessions#new).

Body background does not change

I'm unfortunately no expert, but it seems like if a background on body is set, it does not get changed with Turbolinks

Link is being requested twice. First one is cancelled and second is full-page refresh.

I installed turbolinks and making some tests to see it in action. Although I realized many links are being requested twice. What can I do in order to debug and help this development? I remember this also happening when using pjax. Any tips?

ConstrainPageCacheTo(10) doesn't work when history limit reached

In Chrome, the history limit is apparently 50, and when we reach this limit, this line :
delete pageCache[currentState.position - limit] if currentState.position == window.history.length - 1
doesn't work anymore, so the pageCache grows out of its limit of 10.

Page goes blank with LiveReload in footer

I use LiveReload for instantly updating CSS, etc changes on my Rails sites.
I've just installed turbolinks on one of my Rails apps and the page just shows blank when clicking any links.

When removing the LiveReload script from the footer the site works OK and wow the site is much faster!

page:change event missing

It looks like the page:change event may have been removed by accident.

Removing body also removes event listeners on body

Javascript libraries like Twitter Bootstrap put event listeners on body.

Such as:
https://github.com/twitter/bootstrap/blob/master/js/bootstrap-alert.js
https://github.com/twitter/bootstrap/blob/master/js/bootstrap-modal.js

By replacing body, these event listeners are removed and events aren't handled after a page is changed.

Replacing the body's innerHTML with the replacement body's innerHTML has the correct functionality in these situations. But then you sacrifice cases in which you want to change the body element, its attributes, and listeners.

Perhaps there could be a data attribute on body which configures body replacement for the uncommon case. For example, you could specify data-turbolinks-replaceinner="true" on body and when the page is replaced, it replaces the body's innerHTML and leaves the body element with its listeners.

WDYT?

IE9 JavaScript Error: Invalid Argument

It seems IE9 DOMParser doesn't support "text/html"

jCarouselLite stopped working after implementing Turbolinks

I have a slider of images with jCarouselLite, this stopped working after adding Turbolinks to the project.

All though this is the only page where the jCarouselLite slider isnt working:

$('#employees_slider').jCarouselLite({
      auto: 0,
      visible: 4,
      start: 0,
      speed: 2000,
      btnNext: ".right",
      btnPrev: ".left"
    });

= link_to (image_tag 'left.png', :class => 'left', :width => "48"), ""
= link_to (image_tag 'right.png', :class => 'right', :width => "48", :style => "float:right;"), ""
#employees_slider
  %ul.employee_images
    - @employees.each do |employee|
      %li{:style => "width:200px !important;margin-right: 25px !important;"}
        - if !employee.remote_image_url.nil?
          = image_tag employee.remote_image_url, :id => "employee_#{employee.id}", :class => "employee_image"
        - else
          = image_tag employee.image, :id => "employee_#{employee.id}", :class => "employee_image"
        %p.employee_info{:id => "info_#{employee.id}"}
          = employee.name
          %br/
          = employee.title
          %br/
          %b=t ".email"
          = employee.email
          %br/
          %b=t ".phone"
          = employee.phone

reset scroll position on page replacement

It's really annoying then you visit a link with shorter content and being greeted with a blank page, having to scroll up to see the content. I just experimented with resetting the scroll position to 0,0 on successful fetch and it feels much more natural:

xhr.onload  = ->
    changePage extractTitleAndContent(xhr.responseText)...
    reflectRedirectedUrl xhr
    resetScrollPosition()
    triggerEvent 'page:load'

resetScrollPosition = ->
  window.scrollTo 0, 0

Reload the page when detecting a change in assets

Turbolinks should keep an index of external assets from the first page hit. Imagine these two assets declared in the HEAD:

<link  href="https://d3dma7a8zw52bz.cloudfront.net/assets/application-9ce82b8e23611aebdd79033e8063d82f.css" media="all" rel="stylesheet" type="text/css" />
<script  src="https://d3dma7a8zw52bz.cloudfront.net/assets/application-983dc7ef267ab91bed1488896e7a6f6c.js" type="text/javascript"></script>

Turbolinks should extract the two urls and then compare each subsequent request against them. If there's a change in the URL, it should do a full page reload instead of just turbolinking, such that the new CSS or JS can be reflected.

extractAssets should ignore certain types?

Hi - I've been using Pjax in a limited scope for a while, but only just started trying out Turbolinks in our app, so forgive me if I'm missing something.

I'm struggling with the assetsChanged method always returning true for pretty much any two pages on our site, which then results in Turbolinks doing a full page reload. We have a number of oembed <link> tags in our pages, as well as specialised RSS <links> for differing areas of the site, and some legacy Facebook <link> tags like rel=image_src.

I'm wondering if it would make sense to be able to tell Turbolinks to ignore link tags with certain type or rel attributes (eg "application/xml+oembed"). There are obvious problems with that approach - eg the ombed tags would still refer to an old page, but since they're intended for machine consumption only perhaps it doesn't matter. (RSS is a bit trickier... ideally you might replace the old RSS tags in the head with the new ones, though I'm not even sure if browsers would notice the change. Maybe that doesn't matter though, since having an RSS button in your browser chrome seems to be rapidly disappearing.)

Any thoughts? Or am I doing it wrong?

asset caching doesn't work with async loaded js (like google analytics)

I just updated to the latest and I started getting full page refreshes, I tracked this down to the assetsChanged function which is always returning true because of the async js I'm loading which is present in the cached document but not present in the newly fetched version.

This is breaking for the YouTube API and Google Analytics at the moment. Trying to see if I can find a solution..

Integration problem with Stripe Button

I'm trying to integrate the new Stripe Button as described here : https://stripe.com/docs/button

The button shows up fine on a fresh page load but doesn't show at all when the page is loaded with Turbolink. here is the exact code :

<form action="/charge" method="post">
  <script src="https://button.stripe.com/v1/button.js" class="stripe-button" data-key="YOUR_KEY"></script>
</form>

The JS file has no initialization method. It just executes when it's loaded. Is that a use case that should work with Turbolinks? I see inline JS should load fine but I'm not sure about external scripts.

should ignore links with data-method="delete"

Links with data-method="delete" are resulting in a GET and a DELETE at the moment. If i add data-no-turbolink it does the right thing but it seems unnecessary.

Integration with fancyBox (v1.3.4)

Hi,

I'm having issues with fancyBox integration.

When I reload gallery page everything works as expected.

When I go to gallery page using turbolink the image preview modal is not working.

I'm not getting any js errors.

Here is my application.js

document.addEventListener("page:load", function(){
  init('load');
});

document.addEventListener("page:change", function(){
  init('change');
});

$(document).ready(function() {
  init('ready');
});

function init(msg) {
  $("a.fancybox").fancybox();
  alert(msg);
}

Thanks in advance,
Marek

page:load on first load or back

I'm probably missing something, but it does not seem that page:load is triggered on reload or back (or on first load) of a page.
Is this what is supposed to happen? And if so, what is the best way to catch these events using page:load?

Thanks!

Code used in application.js:

document.addEventListener("page:load", function(){
  alert("loaded");
})

configurable cache element

I've just hacked together a quick implementation of specifying a configurable cache element, so you can do something like this:

Turbolinks.setup({cache_id: 'main_container'});

For instances where you don't want to refresh the entire body. I have this requirement on two projects I'm working on so thought it might be a common case. Thoughts?

link_to helper doesn't work with :back

<%= link_to "Cancel", :back %> generates a link to the current page

skip turbolinks for links where data-remote is set to true

Just stopped working, Google Maps set assetsChanged to true

Today my turbolinks just stopped working. Everything worked yesterday and I have not done any updates. Turbolinks do not even work when i disable all my other javascript files (I use Rails 3.2.8 and have commented out everything exept jquery, jquery_ujs and turbolinks in my application.js).
I have tried to trigger turbolinks manually with Turbolinks.visit(path) but it is the same.

Now when i click on a link, the browser makes a GET request with the new page like it's expected to do but the get response contains the full page and not only the body. After the get request are done, browser do a full page reload.

I have set a brek point in firebug (line 14) to test if the broswer support pushstate and it does, breakpoint never triggered.

I have uninstalled the gem, cleared catch, tested in FireFox 16.0.1, Chrome 22.0.1 and Safari 6.0.1, everything are the same!

Please, help me to solve this really annoying problem. I haven't done any changes since yesterday and then everything worked fine.

Edit,
I installed Oh My ZSH yesterday but guess it does not matter.

How will this handle google analytics code?

Just a thought, how does google analytics behave to this? Does the analytics snippet need changed in order to work?

Or do we handle this another way?

Page transitions

Any chance? Like slides, fades and other fancy stuff?

Firefox 15 Ubuntu.

Works fine in Chrome on Ubuntu. Firefox 15 on Ubuntu returns a white screen, you can reload the page to get the requested page.

Things that need a bit of clarification…

So following the idea from the readme: "So if you wanted to have a client-side spinner, you could listen for page:fetch to start it and page:change to stop it."

Probably startSpinner() would insert some element ("Loading…") into the body. However as on page:change, the body has been replaced, so the inserted object would have been gone. I do not see an easy and obvious way to gracefully end the spinner, it would just disappear.

Something else:
For action specific javascript, I usually have a siteName.controllerName.actionName js namespace and would call its init() method from the view.
How would I translate that to turbolinks? I have not the slightest idea, what controller or action is served to me (if not by some hack, where I would place some

in my body and parse that).

Problems integrating Turbolinks with Facebook JS SDK

This is not really a Turbolinks issue but a problem I guess many will run into using the Facebook JS SDK. The SDK is initiliazed inside the HTML body like this: https://developers.facebook.com/docs/reference/javascript/

When the SDK initializes, it creates a fb-root iframe tag to communicate with the FB API. When Turbolinks reloads the page, the fb-root tag vanishes and there doesn't seem to be a way trigger the re-initilialization of the Facebook sdk.

I'm trying to figure out the best way to handle this. Can an element of the body remain in between two requests? How can inline javascript be executed? (Let me know I should repost to Stack Overflow instead)

Cache issue when mixing 'back' button an Turbolink visit

Found an odd bug but I can't seem to track down how to fix the problem.

Steps to reproduce:

1. Start at http://localhost:9292/test/index.html
2. Click the Other page link.
3. Hit the back button on the browser.
4. Click the Other page link.
5. Click the Home link.
6. Hit the back button on the browser.

You'd expect to see the other.html page which is a single link, but nothing changes and instead you see the content for index.html

Events observed in script tags are still observed when changing page

Related to #58 (JS script tags of pages visited with turbolinks are executed).

If you observe events in one of these script tags, when you go to another page, the event handler is not removed.
It's problematic, for different reasons, one of them is that if you come back to this page, the event handler are executed twice (or three times if you come back another time, etc.)

jQuery UI tips and tricks

Just leaving some notes for some things i noticed when using jQuery ui components (particularly the datepicker)..

When a page has changed, the .datepicker has not reinitialized because it doesnt check to see if its datePicker div has been removed from the dom, but it has.

To correct this you can do something like this...

jQuery.datepicker.dpDiv.appendTo( jQuery('body') );

I updated this solution per @rpflorence ... thanks for the catch.

Lost Anchor #

Turbolinks seems to loose the anchor on page:change.

Given a link href="/foo/bar#100" it will show this link in the URL until page:change is triggered.

After that it's just "/foo/bar".

Could someone confirm this behaviour?