- Added the following notifications to $.mobile.loadPage():
- pagebeforeload
- Triggered just before loadPage() attempts to dynamically load an external page.
- Developers can prevent the default loading behavior by calling preventDefault() on the event. If preventDefault() is called, it is up to the developer to call resolve()/reject() on the deferred object passed within the data object (2nd arg to the event callback).
- pageload
- Triggered after an external page has been loaded and inserted into the document.
- pageloadfailed
- Triggered when the load of an external page fails.
- Developers can prevent the default behavior (error dialog display) by calling preventDefault() on the event. If preventDefault() is called, it is up to the developer to call resolve()/reject() on the deferred object (2nd arg to the event callback).
scrolltop as a solution isn't that great but some browsers scroll to the top
of the page to where the element bearing the id matching the hash is located
*before* the hashchange event is fired meaning we don't have an opportunity
in the changepage event lifecycle to record the scrolling properly
- Modified loadPage() so that if the data-url lookup for a given page fails, that it look for the page via id (if it is an embedded page URL). This allows us to find dynamically injected pages that are un-enhanced and missing their data-url attributes.
Dialog not working if $.mobile.ajaxEnabled = false
https://github.com/jquery/jquery-mobile/issues/2451
related/dup bug 2202:
Dialog loads in new page with ajaxEnabled = false
https://github.com/jquery/jquery-mobile/issues/2202
- Modified the default click handler to check if the href is for an embedded page before bailing when ajaxEnabled = false. This allows us to navigate to internal/embedded pages/dialogs on the click versus waiting for the accidental hashchange that was the result of the browser's default handling of hash fragments.
- Modified the url parser regexp so that we can find the double slash that precedes the authority. This is necessary so we can reconstruct resource urls used on some devices like Rim's Playbook that use urls like:
location:/dir1/dir2/file.html
- Modified makeAbsoluteUrl() so that it uses the new doubleSlash property in the object returned from parseUrl() instead of assuming that it is ok to use a double slash.
- Make sure that our hashchange resolves non-path hashes against the documentBase. This prevents the resulting changePath() call from incorrectly resolving against the URL for the current active (external) page.
- Fixed a problem in the push-state code. A hashchange event is *NOT* fired when navigating back (window.history.back()) from an external page to an internal page. This makes sense when you think about it since hashchange is only ever fired when the hash of the current document url changes, not when the document url itself changes. The fix was to make sure that the pushstate hashchange callback always sets a state object, even on embedded page URLs. This allows the hashchange callback to be triggered from within onPopState().
Also checking in the first example of how to use the pagebeforechange notification to allow for dynamically updating and re-using a page that is already in the DOM.
- Added "dataUrl" option to changePage(). This allows a caller to specify a page element to change to, but specify an alternate URL for location display purposes. This is useful for dynamic applications that re-use and over-write existing page content to avoid overwhelming the DOM.
- Renamed the "beforechangepage" and "changepage" events to "pagebeforechange" and "pagechange" respectively. This was done to match the page widget naming of its notifications. Left the triggers for the old events in place but with DEPRECATED comments.
- Renamed the properties of the data object passed to the page events.
- Moved the setting of isPageTransitioning *AFTER* the beforechangepage notification.
- Modified the trigger("beforechangepage") call to pass the args to changePage() as an object since trigger only expects one data arg.
Also, this change removes a reference in memory that we were keeping to the $activeClickedLink on each page. We stored this in attempt to refocus a link after returning to a page. Unfortunately, it doesn't seem that this data can be retained after pages are removed from the DOM, outside of somehow remembering a unique selector string to reach that element again (which could be achieved by adding some overhead, ala http://stackoverflow.com/questions/2068272/getting-a-jquery-selector-for-an-element )
Turns out this little code was enough to invoke the error:
"/dir1/dir2".replace(/\/?/, "");
Rewrote the regexp in path.makePathAbsolute() that was stripping leading slash, and trailing filename/slash. This gets around the problem.
Special thanks to @adambiggs for helping me test 33 iterations when trying to narrow down what in jquery.mobile.navigation.js was choking BB5.
- Added code to calculate whether to choose the documentUrl or the page Url in the case where an action is not specified on a form element.
- Fixed bug in the navigation "submit" handler where an error was being thrown if "type" was not specified.
- Fixed typo "diabled" id typos in tests/unit/navigation/index.html.
- Added tests for form submissions with no action in both base and non-base specified cases.
- Don't set a default value for the 'role' option to loadPage() and changePage(). Instead, rely on the role attribute that is on the element.
- Fixed an error that was happening when a hash change involves going from a dialog back to another dialog. I had recently made changes that assumed the 'to' variable was always going to be a url string, but it turns out in the dialog back to dialog case, it is an actual jQuery collection.
- Added changepagefailed notification for the case where the load fails. Note we still need beforeloadpage/loadpage notifications.
- Misc fixes to stop the navigation unit tests from seemingly random failures. It seems that the hash resets in setup/teardown of the test modules sometimes interferes with the changePage() requests in the tests. Added code to the setup/teardown functions to wait for the reset to complete before moving on to the tests.
