YUI 3 :: {126} Tickets Addressed in 3.6.0

Allow Selecting Dates In Calendar Before Rendering

You are unable to call the .selectedDates() method of a calendar widget before rendering the calendar. It would be nice to be able to use this method before calling the .render() method. I think it would be even better to be able to specify the selectedDates from the constructor of the widget like so:

new Y.Calendar({ selectedDates: myDates })

Here is a fiddle that can be used. Move the .selectedDates() call above .render() and you'll get an error in the console.

http://jsfiddle.net/ddelrio1986/Kev8J/2/

Calendar Rules overhaul

- Instead of reparsing strings every time a rule is matched, generate closures for each rule and apply them in succession.
- Support date ranges as a top-level definition for the rule sets.
- Support for "nth weekday of the month"

Hardcoded h4 in HEADER_TEMPLATE

Line 1005 and 1028 of calendar-base.js have the H4 from the HEADER_TEMPLATE hardcoded:

contentBox.one("." + CAL_HD).one("h4")

This makes the HEADER_TEMPLATE less customizable as the H4 needs to be there or else the afterChange events fail.

Calendar Day Template breaks click event

I overwrote the CALDAY_TEMPLATE to contain a .liner div around the day number. The _clickCalendar callback is using ev.target, the clickedCell is getting assigned to div.liner, and not the td. Changing clickedCell = ev.target to clickedCell = ev.currentTarget fixed this for me.

Example tests for Calendar

Implement example tests for Calendar

hover style not cleared when navigating with keyboard

If I click on a date, the hover style is also active (dark blue). If I then move the day with the keyboard and change to another month, the hover style is still active on the date that was originally hovered. E.g. if I click on the 17th, then move to the next month via keyboard, the 17th of the next month has the hover style (because it's the same <td>, obviously).

The hover style should be cleared on the first keyboard interaction.

Build out calendar unit tests to achieve 100% code coverage

Refocus on calendar nav elements after they are activated via keyboard in IE

See summary

Rules object resets if rules are specified at render time

See summary.

No selection rendering is bound to selection change event.

See summary

Removal of dates from selection doesn't function correctly if removeDates private method is used.

See summary.

CustomRenderer attribute should be lazyAdded

See summary

setYear is deprecated

According to MDN (see URL), the setYear method on the Date object is deprecated in favor of setFullYear.

Example tests for Date and Number

Implement example tests for Date and Number

Deprecate Y.DataFormat.Date/Number in favor of Y.Date/Number

See summary

SWF utility doesn't allow for preventing autoplay and loop

Please see http://yuilibrary.com/forum/viewtopic.php?p=28211. In there I mentioned that the SWF does not support the fixedattributes loop and play. Without the ability to set loop="false" and play="false", the movie file starts as soon as the page loads and loops continously. I know these properties do not apply to Flex but Flash applications should support them. Please add the 'loop' and 'play' to the SWF's fixed property list.

Thanks,
Sridhar

ExternalInterface.addCallback does not work with Y.SWF, Flash 10.0 and IE // Test and incorporate Tripp's patch

Methods added via ExternalInterface.addCallback do not work in Flash 10.0 and IE when embedded with Y.SWF.

Replicated in Flash 10,0,22,87.

Embedding the same swf with swfobject 2 works as intended.

Source files attached as a ZIP.

Useful links:
Archived Flash Player versions: http://kb2.adobe.com/cps/142/tn_14266.html
Remove Flash Registry Key to install older version: http://kb2.adobe.com/cps/116/4da116d3.html

Review TreeView code and example

See summary

Standardize uploader event names

Uploader publishes "uploaderReady" but also passes through all-lowercase event names from the SWF, resulting in mixed formatting event names and even mixed prefixes. These event names should be updated before being relayed.

Uploader should also be using Y.Base's NAME property to auto-prefix capability.

For example, these events:
uploaderReady
uploadcomplete
uploaderror

Should be:
uploader:ready
uploader:complete
uploader:error

So that implementers can use uploader.on('error', ...) syntax.

More Robust File FIlters

Only basic allowed extension file filters are supported in the Flash Uploader, this does not seem to be supported in the HTML5 uploader. Wondering if this feature is being abandoned or if it is on the roadmap for the HTML5 Uploader as well. If so I think it would make good sense to also offer disallowed extensions, and also a max/min file size. I find I am building this functionality now in my implementation but it occured to me that this may be something that is desirable for all.

File List Manipulation

There is no mechanism to control and manipulate the built in (multiple) file list that the uploader control uses. You can add files, but you cannot remove them. It is pretty easy to add an event to clear the list entirely, but I am finding it much more challenging to remove individual files from the list using the out of the box solution. This also makes it difficult to perform validations (i.e. file size, file extension, max # of files) before adding a file to the list. I am going to build a solution for my implementation to manipulate an array of Y.File objects to allow for these adjustments and use UploadThese to send them to our backend.

HTML5 Uploader: Drag 'n Drop Functionality: Drop event is not being fired

When a file is "dropped" in the drop area the "drop" event is suppose to fire.

However, the "this.fire("drop");" is missing from the source code. In uploader-html5.js between lines 335 thru 351. The drop event should be fired.

Can't upload the same file twice with webkit

Hi,

When you try to upload a file twice within webkit it doesn't work:

You can test by yourself: http://jumbay.info/yui/uploader/

Select a file, upload it, then select the same file, the "fileselect" event won't fire.

Uploader Flash, empty fileList array after fileselect event

After fileselect event fileList array is empty for a few milliseconds. uploader.uploadAll() in fileselect event callback function is impossible.

Set a XMLHttpRequest header in the request header.

I am using uploader in a progressive enhancement scenario.

Normally a file input and a submit button is displayed. This is then hidden and replaced by the uploader's buttons. Normally, without javascript, the page will refresh and be rerendered. However, with the uploader, depending on the upload, I would just like to send back a simple JSON response.

Normally, with Y.IO, I can check if XMLHttpRequest is the value of the X-Requested-With header to determine if it's an AJAX request. With Uploader, however, the X-Requested-With is not set, which results in a not so neat solution:

Setting an extra POST variable, for example, called "isAJAX", and then check that on the server side.

It would be much neater if the X-Requested-With header is set, so that we can determine if the request is "AJAX" or just an ordinary request.

Uploader: IE10 does not upload files

IE10 with the "Multiple Files Upload with Transparent Overlay Button" only seems to partially finish the file upload and displays "Hasn't started" for the first file.

Example:

File name | File size | Percent uploaded

lol.txt | 3 | Hasn't started yet
rofl.txt | 4 | 100%

Total uploaded: 100%

Uploader: IE10 has no server data retrieved

Example "Multiple Files Uploader with POST Variables and Server Data Retrieval" has no data from server received in IE10.

Minor error with uploader's documentation

I found a small error with uploader's documentation (features based on avaliable functionality) here: http://yuilibrary.com/yui/docs/uploader/#features-based-on-available-functionality

I believe this code here:



uploader.set("fileFilters", [{description:"Images", extensions:"*.jpg;*.png;*.gif"},

                                    {description:"Videos", extensions:"*.avi;*.mov;*.mpg"}];);

Should be:



uploader.set("fileFilters", [{description:"Images", extensions:"*.jpg;*.png;*.gif"},

                                    {description:"Videos", extensions:"*.avi;*.mov;*.mpg"}]); //removed extra semi colon.

Explicitly state necessity of Flash on Uploader Examples for IE

Testing Uploader on IE without Flash installed gives the message:

"We are sorry, but the uploader technology is not supported on this platform."

This should be changed so that the user understands that Flash is required.

All Uploader examples fail in Chrome on Android 4.1

Credentials issue in Uploader examples in FF14

withCredentials changed the default value, which causes issue in Uploader examples.

Pull in existing Uploader unit tests

Flash Uploader does not fire error events correctly

https://github.com/yui/yui3/blob/master/src/uploader/as/FlashUploader.as#L391

Looks like there's a missing event.type assignment, which results in events not getting correctly relayed to JS.

A patch SWF will be included in this ticket once it is produced.

Uploader Queue does not push the next file up to upload when a running upload is cancelled

The cancelUpload() method is not starting the next file once a running file is cancelled.

Uploader issue in Safari because withCredentials is set before XHR is opened

See summary

Deprecated modules should be excluded from "Related Examples" box on the right

Uploader examples: Related Examples box on the right points to deprecated examples, but we shouldn't link to those from there.

contentready event documentation has two weird selectors

In the example documentation at the URL listed, it looks like the ID selectors inside the javascript are incorrect. There are 3 different IDs which I think are all trying to refer to the same element:

var notHereYet = Y.one('#list'); // null

Y.on('contentready', addItem, '#todo');

error in documentation for CSS buttons

on this page: http://yuilibrary.com/yui/docs/button/cssbutton.html

Source: HTML

<link ref="stylsheet" href="http://yui.yahooapis.com/3.5.0/build/cssbutton/cssbutton.css">

S/B link rel= not ref='

Additional tests for Widget-StdMod

Tests for setStdModContent() and the HTML_PARSER.

Error in documentation for Handlebars

In the "Partials" section,

The line :
Y.Handlebars.registerPartial('header', '<h1>Handlebars</h1>');

should actually be:
Y.Handlebars.registerPartial('header', '<h1>{{title}}</h1>');

This shows the inheritance of the context in partials.

In the example, the partial template should renders to "<h1>My Page About Mustaches</h1>"
With the provided code, it renders to "<h1>Handlebars</h1>"

Improper use of "it's" in docs

"It's" is not the possessive form of "it," it is the contraction of "it is."

The following:

A YUI module can be described as "any code that can be separated to run on it's own". Many times, this code can be reusable in different ways.

Should read:

A YUI module can be described as "any code that can be separated to run on its own". Many times, this code can be reusable in different ways.

This would have been a pull-request but your tech writers haven't committed the web site to git yet...

Drag handles aren't well explained in the example.

The example here: http://stage.yuilibrary.com/yui/docs/dd/drag-plugin.html

It demonstrates using the dd node plugin. The drag handle is an h2 element that says Drag Me Here. That's a good visual clue. The old example has a little x in the top corner to drag by, and that was confusing since it appeared to be a close button. The new example failed to change this comment in the source: //Now you can only drag it from the x in the corner

Also, I think it's pretty obvious what happening by looking at the example, yet I feel like there should be at least a 1 sentence description of what a drag handle is and does.

DDConstrained does not remove event listeners on destroy properly

Even if we unplug DDConstrained plugin, the node is still constrained, because the listeners were not removed.

DD Delegate ignores dragNode configuration option

DD Delegate allows you to pass in a configuration object, dragConfig, which is later used when creating the Drag instance. However during mousedown event handling the 'dragNode' configuration option is overridden without checking the 'dragConfig' object first.

in _delMouseDown



if (dd.proxy) {

    dd.set('dragNode', Y.DD.DDM._proxy);

} else {

    dd.set('dragNode', tar);

}

instead it should be something like



if (dd.proxy) {

    dd.set('dragNode', Y.DD.DDM._proxy);

} else if ( conf.dragNode ) {

    dd.set('dragNode', conf.dragNode);

}else {

    dd.set('dragNode', tar);

}

Drag and Drop Window scrolling not working when dragged vertically

I am testing using this example here: http://yuilibrary.com/yui/docs/dd/winscroll.html

When dragging the div horizontally, the window scrolls as expected.

When dragging the div vertically, the window never scrolls. This is regardless of whether the browser is maximized or whether it has been resized to take up part of the screen.

This is happening in all browsers.

Y.Plugin.Drag interferes with browser's native handling of dragged links, even when those links aren't targets

It's usually possible to drag a link into the bookmarks bar in a browser to bookmark that link directly. However, if Y.Drag or Y.Plugin.Drag are used, it's no longer possible to drag a link and have the browser handle it as normal -- even if the node is not a target, or explicitly marked as Invalid.

See the attached fiddle for an example. I've just added the addInvalid too - but it doesn't work either with, or without it.

bz #5540030 - Indent/Outdent button doestn't work properly in Chrome/IE when in RTL mode

While Indent/Outdent works properly with LTR mode, it doesn't work with RTL
mode in Chrome/IE. It works properly in FF.

nodeChange event's command key loses both the underline and strikethrough property if both are applied to text in the text editor

If you underline and strikethrough text in the texteditor, nodeChange events connected to the text will not have underline or strikethrough in the commands object in the event object. Only when you apply both underline and strikethrough to the text does the commands object lose both aforementioned commands.

page not found

while browsing the IO API, I received the following error

Not Found

The requested URL /yui/docs/api/modules/io-nodejs.html was not found.

Y.io in nodejs fails when config.data is an object

config.data needs to be manually stringified

loader docs specify incorrect maxlengthurl

docs say 2048, code says 1024

not sure if this is a mistake in the docs or the code, but they don't match up.

The trigger field under modules definition should support multiple module names

I could have this yui config: {"modules":{"mymodule":"condition":{"trigger":"dom-base","test":function(){return true;}}}}
I think trigger should accept an array of module names so if any of these modules are being used, mymodule would be used.
btw - the condition field is not documented under http://developer.yahoo.com/yui/3/yui/#loader.

Configuring the loader for modules that ship with only the root bundle.

If I want to write a module that will uses Y.Intl bundles for strings, I would typically start with a root bundle, and then add language specifics later. To configure the Loader to load bundles for a module where only the root bundle is currently defined is a bit inconsistent. I actually have to provide an empty string in the array for the lang property {lang:[""]}, something which I don't have to do when I have language specific bundles.

YUI instance's combine:false should override group's combine:true

First, I configure a group:

window.YUI_config =
{
groups:
{
ygallery:
{
ext: true,
combine: true,
base: 'http://localhost:8666/combo?',
comboBase: 'http://localhost:8666/combo?',
root: '',
patterns:
{
'ygallery-': { },
'lang/ygallery-': { },
'ygallerycss-': { type: 'css' }
}
}
}
}

Then I require them:

YUI({
filter: 'debug', combine: false,
gallery: 'gallery-2012.01.11-21-03'
}).use('ygallery-foo', 'ygallery-blah', function(Y) {...});

When I view the generated source in Firefox, I see this script request:

http://localhost:8666/combo?ygallery-foo/3.5/ygallery-foo-debug.js&ygallery-blah/1.5/ygallery-blah-debug.js

I think the combine:false configuration for the YUI instance should override the group's combine:true configuration.

In addition, I think that Loader should fall back on using comboBase even for single modules, if base isn't defined.

Enhance YUI Loader to support module versions

I'm working on a combo handler that understands versions, to allow each module to have an independent update history, and any combination of module versions can be loaded. (This is different from the YUI 3 Gallery, where only a single gallery version can be specified.)

This patch seems to do the trick:

_path = function(dir, file, type, nomin) {

+ if (Y.config.moduleVersions && Y.config.moduleVersions[dir]) {

+ dir += '/' + Y.config.moduleVersions[dir];

+ }

var path = dir + '/' + file;
if (!nomin) {
path += '-min';
}
path += '.' + (type || CSS);

return path;
};

I configure a map of module names to version numbers in YUI_config (along with the module group which defines comboBase, root, and patterns), and the above patch automatically includes the version number in the path sent to the combo handler.

YUI_config = {moduleVersions:{'ygallery-foo':'3.5'}, groups:{...}};
YUI().use('ygallery-foo', function(Y) {...});

This generates a request for

{comboBase}ygallery-foo/3.5/ygallery-foo-min.js

and similarly for any CSS. This directory structure makes it very easy to store many different versions for each module.

Loader may calculate incorrect css file paths for different skins

Requesting a non-native module from two different Loader instances with different skins will calculate incorrect css file paths for all but the first Loader instance that calls resolve.

Full write-up available at http://www.brainsfordinner.com/yui/yui-loader-bug.html

Combo Loader broken in 3.5.0pr4

YUI combo loader is broken in 3.5.0pr4 and fails to load js files in the correct order. But it does look like these are loaded correctly in the Chrome debugger. This is major show stopper.

The test case is below. If you keep pressing F5 in Chrome, you will get 'Uncaught ReferenceError: a is not defined' eventually in Chrome:

ComboTest.html

}
};

return config;
};

YUI(getConfig()).use('b', function (YUI3) {
});
</script>

a.js file:

a = function () {
};

b.js file:

b = function () {
};

b.prototype = new a();

Allow for cascading module requirements

It would be very cool if loader allowed for taking required modules that are shared and specifying those at a higher level as opposed to needing to repeat them for every module.

https://gist.github.com/2259463 has some examples of how I think this would work.

Implementation-wise, it would merge requires from each level together for the final module. In that respect it would be like base/root/etc work now either at the root or for groups. The difference would be unlike these simpler properties the arrays would be concatenated together.

YUI config does not properly override skins

We have a sample of our CONFIG parameter that we pass in here:

http://jsfiddle.net/UYEvU/

We override some skins with an empty array to avoid loading the default skins.

Repro:
Run the above on both 3.4.1 and 3.5.0 and watch the CSS tab of your browser's inspector.

Actual:
3.4.1 obeys the override and does not load widget-base.css or widget-stack.css
3.5.0 does not obey the override and loads http://yui.yahooapis.com/combo?3.5.0/build/widget-base/assets/skins/sam/widget-base.css&3.5.0/build/widget-stack/assets/skins/sam/widget-stack.css

This results in a large performance degradation; our hack for the meantime is to edit the yui/yui.js file directly.

Performance problems on YUI().use on FF3.6 and IE8

We have recently upgraded YUI to its latest version to find that for FF 3.6 and IE8 is much slower to call YUI().use than it was on YUI3.4. We are aware that there is functionality on 3.5 that did not existed on 3.4 but still the perforamnce penalty in those browsers is too high.

The problem comes with the _inspectPage method in the YUI object. This method builds everytime the instance is called the dependency tree, there is no caching whatsoever among instances. As far as I know there are two possible methods to load dependencies in YUI:

1) Use YUI().use('whatever-module')
2) Load JS directly with a "script" tag.

There is no method to remove dependencies, then, it seems that caching of what has been already loaded should happen across YUI instances. If a module is loaded is loaded "for ever", thus why do we need to rebuild the dependency tree everytime from scratch when the YUI instance is found?

Intl.get doesn't load language modules for non-yui modules defined in a group with configFn

While upgrading from 3.4.1 to 3.5.0, this configuration stopped working.
We have external yui modules, pulled in from a private server (no combo handling) and defined as groups (as per http://yuilibrary.com/yui/docs/api/classes/config.html#property_groups). The path of those files is defined in the configFn function.
Some of those modules need to get their language packs through Intl.get.
Those modules are YUI().use'd in html pages where their language packs are written (http://yuilibrary.com/yui/docs/intl/#packaging).
In the attached archive, you'll find an example of this html page; also, a fake module (mygroup-util-1.4), that (given the configFn), will be searched in http://my.server.it/js/util/1.4/util.js.
Note: paste this in the Firebug console and this will work:

YUI({lang: 'it'}).use("mygroup-util-1.4", function(Y) { Y.log(Y.mygroup.test()); })

Loader error on Node Js

I'm running Node v0.6.17 on Fedora 16 with yui@3.5.1.

I'm trying to load a custom module from the current directory. I'm pretty sure that the problem is because the config object I'm passing to the YUI factory method is invalid, but it's not failing in a clean way. I get this:

/home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:6450
var f = this.onFailure, msg = [], i = 0, len = o.errors.length;
^
TypeError: Cannot read property 'length' of undefined
at Object._onFailure (/home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:6450:64)
at Object.<anonymous> (/home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:6625:24)
at [object Object].<anonymous> (/home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:6692:30)
at /home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:3747:26
at /home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:3788:21
at /home/username/node_modules/yui/yui-nodejs/yui-nodejs.js:3725:25
at [object Object].<anonymous> (fs.js:98:14)
at [object Object].emit (events.js:64:17)
at [object Object].<anonymous> (fs.js:92:18)
at [object Object].emit (events.js:67:17)

I tracked it back to this line of code: https://github.com/yui/yui3/blob/master/src/loader/js/loader.js#L1961
It looks like o.errors is definitely expected to exist.

I'll try to fix my config object. I will post updates here if I find out more about what's going wrong.

Loader loads language files for custom modules from YUI CDN

When defining a custom module with languages, the loader tries to get the language files from http://yui.yahooapis.com :

loader.calculate({ignoreRegistered:true}) still pulls in dependencies from a previous run of loader.calculate()

When I call loader.calculate() with ignoreRegistered:true, I'm still getting results from a previous call to loader.calculate(). Specifically, I'm recalculating the dependencies for the same module, but with a different set of modules registered via loader.addGroup().

Skin override loads 2 stylesheets: sam and the custom one

See http://yuilibrary.com/forum/viewtopic.php?f=93&t=9801

Unlike in 3.4.1, if you select a skin override in 3.6.0, it will load both the sam skin and the custom skin. The order in which this happens doesn't seem to be guaranteed at first, which is why I was confused at first because the isolated test seemed to work.

Loader does not always use of configFn

See the attached example code. It is not necessary for the example code to work, the only thing of interest is WHAT PATH the modules are attempted to be loaded.

I have a "group" section and in it a configFn which changes the path, because my directory structure is not "module-name/module-name.js" but just "module-name.js", with all JS files in one top level build/ directory.

When I don't include the "module:" information everything works. When I DO include the module information the following strange thing happens:

/build/part1-xhr-datasource-debug.js"
/build/part1-util-debug.js"
/build/part1-init-configevent-provider-debug.js"
/build/part1-settings-ds-provider-debug.js"
/build/part1-busyloader/part1-busyloader-debug.js"
/build/part1-textnode/part1-textnode-debug.js"
/build/part1-treeview-debug.js"
/build/part1-treeview-json/part1-treeview-json-debug.js"
/build/part1-test/part1-test-debug.js"
/build/part1-base-debug.js"

As you can see SOME files are attempted to be loaded USING the configFn, and some DON'T use the configFn, attempting to load the file from "/build/module-name/module-name.js".

if Y.Loader is loaded (without meta), it should allow to load modules defined in meta

With the new loader.resolve(true) we can generate (pre-compute) all the metadata needed for the app instead of loading the whole meta from CDN (loader-min.js) to improve performance. The problem with 3.5.x is that it will always try to load "loader-min.js" no matter what.

Here is an example, loading a custom module without any YUI dependency:

YUI.applyConfig({
"fetchCSS": false,
"modules": {
"foo": {
"fullpath": "http://caridy.github.com/examples/blog/foo.js",
"name": "foo",
"type": "js",
"requires": []
}
}
});
YUI().use('foo');

If the seed file is:
http://yui.yahooapis.com/combo?3.5.1-2/build/yui-base/yui-base-debug.js&3.5.1-2/build/loader-base/loader-base-debug.js

then the loader meta doesn't need to be loaded.

Update default Selleck Templates

Update the default Selleck templates that we use to include a querystring parameter that will turn on a predefined test module to test the content of the example for CI integration.

Example:

foo.mustache //example
assets/foo-tests.js //This will contain a YUI module named ('foo-tests')

The main page will contain logic to include a base test module that will include this module, the test module and the test console and call TestRunner.run();

Sortable - JS error when drug between two full join lists after manual nodes manipulation

I create two Sortable lists with full join http://yuilibrary.com/yui/docs/sortable/sortable-multi-full.html

If manually remove or add some nodes into lists it stops work even after sortable.sync()

Please loook on following example: http://jsfiddle.net/pqr7/A23WT/

1. Drag one element from List 2 into List 1
2. Click on "Recreate List 1" - List 1 will be filled with new elements
3. Try to drag one more element from List 2 into List 1 => JavaScript error (in Firefox: "x is null", in IE 9: "Unable to get value of the property 'style': object is null or undefined")

Would like to see a context property in the YQL module

Right now if you want the callback to operate in a different context you have to; send it as a param, bind the function or use a reference variable. I would like to see a context property much like other callbacks in YUI.

Misspelled word on Add Behaviors to Objects example

"extention" should be "extension"

YUI configuration to delay use() callback until domready or window load

Depending on where implementers put the YUI instantiation code, there will be a race between the async module loading and the DOM tree assembly status. In YUI 2, it was typical to use YAHOO.util.Event.onDOMReady(...). In YUI 3, the domready might occur before or after the used modules are loaded and applied to the Y instance.

It appears the only way to ensure safe DOM manipulation would be to either put the instantiation code at the bottom of the body or wrap a decent chunk of the use() callback in a Y.on('domready', ... ) subscription.

It would be convenient to have this eventuality handled by a constructor config, e.g. delayUntil: 'domready' or 'window.load', whereby the use callback would be wrapped in the appropriate event subscription under the hood.

I can also imagine something like delayUntil: '#container.contentready', but the selector syntax vs the target and event identification would be messy (and require Selector).

Y[UI()].use() callback's 2nd parameter: 2 issues

When performing a Y.use(modules, callback) the callback gets TWO parameters: The YUI instance, AND a response object from the loader.

Issues:

- That 2nd parameter is not documented.

- The response is "success" when I try to load a module that the loader does not know about:

Y.use('no-such-module', function (Y, result) {console.dir(result);})

Hoowever, when I have a group-configuration for my own modules and try to load a non-existing module matching that group's pattern the loader goes out and trys to load it, and in this case I DO get an error response.

In general, I think having the response object is MOST helpful. I know in the YUI-configuration I can specify onsuccess and onfailure handlers, but the disadvantage of them is they are global and I would like to know about failures locally. I could set locally-aware functions before each use of Y.use(), but that's a lot of work compared to simply using the response status inside the callback function. If one loads all modules upfront the global handlers are fine, but when one has a lot of loading going on on demand locally in many places in the code it would be so nice to just use that response object.

So I would like to vote for making it "offical" by adding it to the documentation, and for making it work consistently (returning an error in the above example where there is no XHR call going on because the loader decides right away there is no such module).

div is appended precedent to node

While debugging i came across a div (#yui3-css-stamp) which is slightly out of place afaiac.

It is inserted by the following lines, which you can find at yui-core.js at line 451.

Could the stamp div be placed inside the body element?

Module/use callback load errors do not provide useful stacktrace

The issue is that the Y.use callbacks and the dependent module function errors are reported with a stack trace that doesn't contain the problematic file:line reference.

ERROR: use callback error
INSTANCE.log yui-log.js:71
INSTANCE.message yui-log.js:111
proto.error yui-core.js:1155
proto._notify yui-core.js:826

Whereas it should provide a stack trace which contains the reference to the problematic line (Button.html:201 in this example):

Uncaught TypeError: Object #<Object> has no method 'puts'
(anonymous function) Button.html:201
proto._notify yui-core.js:819

I am marking this as a defect because YUI is a developer library and as such one of the major focus topics should be dev time improvements.

Y.Parallel should push arguments if fn is not specified

The way it is now, if I want to save the results passed to the callback, I have to pass a function which echoes its arguments. There is great value in providing the option to transform the results before pushing them onto the result stack, but the default should be to push the raw results, not to push a blank.

Missing requirements sometimes return in wrong order

Reference: http://yuilibrary.com/forum/viewtopic.php?p=32643#p32643

Create a button README

Button appears to be missing a README.md

Basic button widgets example page name change

Basic Button Widget's page name is conflicting with the index page when testing

ButtonGroup incorrectly handles Buttons that contain innerHTML

My button group was being created with HTML similar to the following:

[code]<div class='mode-group'>
<button class='list-view-mode'><div class='yui3-button-label yui3-button-icon'>List View</div></button>
<button class='grid-view-mode'><div class='yui3-button-label yui3-button-icon'>Grid View</div></button>
</div>[/code]

I was then creating a ButtonGroup on the mode-group node. However, when clicking on the buttons the selected state is incorrectly applied to the label nodes unless I'm very careful and click on the button node itself. Thus selection state gets all messed up.

I had a look through button-group.js and I think the error occurs in _handleClick around like 126:



            if (!isSelected) {

                buttons = group.getButtons(); // Todo: getSelectedButtons()? Need it to return an arraylist then.

                buttons.removeClass(selectedClass);

                clickedNode.addClass(selectedClass);

                group.fire('selectionChange', {originEvent: e});

            }

clickedNode in this case is the target of the click event, which in the case of a label using the above HTML is not the correct node to add the selectedClass to.

Unnamed unit tests in console-filters

There may be some unnamed tests in console-filters. Please name them so they can surface in our automated tests properly.

ScrollView can't scroll sprint EVO running android 2.2

I have a project that is utilizing ScrollView that will support the iOS and android. ScrollView works great in iOS, but is a little choppy in android 2.2. The biggest problem is with the sprint EVO, you can't scroll at all. I tried the example from the library and get the same result.

dynamic Update the content of a scrollview

I use a scrollview to show the messages. The message will be updated automaticly by use of a timer.
under the firefox the scrollview works well.
With ie, the content cannot be pulled down to the bottom.

the code clip is as follows:

//root is the result of a webservice return.
var root = o.responseXML.documentElement.firstChild.nodeValue;
var re;
re = root.split("&@&");
Y.one("#lmc").set("value", re[0]);
if (re[1] != "") {
Y.one("#OrgMsg").set("innerHTML", re[1] + Y.one("#OrgMsg").get("innerHTML"));
if (msgtag == 0) {
msgtag = 1;
var scrollView = new Y.ScrollView({
srcNode: "#orgmsgslide",
height: 450
});
scrollView.render();
}
}

ScrollView example tests

ScrollView needs functional tests on the examples. Stubs to start, then fill in later.

[Pull Request] - Allow nesting of scrollviews by stopping gesture propagation

Nesting of scrollviews almost works, except gesture events propagating up the tree cause flicking the inner scrollview to flick the outer scrollview as well (doubtfully a desirable consequence, especially if both scrollviews are oriented on the same axis). stopPropagation() fixes that, although it is a little aggressive, would be nice to be able to declare event listeners as belonging to a certain group, that could be stopped selectively.

ScrollView bounds twice

Message

I'd like to fix the probrem of ScrollView in 3.4.1.

The content bounds twice when I flick it from the edge beyond the end.
Particuraly this behavior can be observed when both the parameters "bounce" and "deceleration" are set near "1".

It seems that the cause of this probrem is that the handler of gesturemoveend event conflicts with that of flick event.
However the position of the content is set to this edge in _onGestureMoveEnd(),
it is set continually in _flickFrame() until "velocity" is over the threshold(=0.015).

I tried to fix this problem like following statement, for your information.

$ diff -u scrollview-base{,.new}.js
--- scrollview-base.js 2011-11-21 11:10:54.000000000 +0900

+++ scrollview-base.new.js 2011-11-21 17:25:23.000000000 +0900

@@ -740,6 +740,12 @@
vel = sv._currentVelocity,
step = ScrollView.FRAME_STEP;

+ if (sv._snapToEdge) {

+ sv._flicking = false;

+ sv._killTimer(false);

+ return;

+ }

+
if (scrollsVertical) {
maxY = sv._maxScrollY;
minY = sv._minScrollY;

Thanks,
Yuta Igarashi yuuigara@yahoo-corp.jp
Yahoo! JAPAN

Scrollview build creates extra css files

The Scrollview build creates a bunch of extra CSS files that are not used. It causes confusion to those that need to build ScrollView.

Horizontal scroll indicator rests out of bounds to the right

If you flick to the last image (the right-most), the scroll indicator's rounded end is obscured by the bounding box. This is different than the visual treatment when the scroll indicator is all the way to the left.

Improve optimizeMemory feature

It currently is a little buggy. For example, when you continue scrolling from one page to the next without letting the scrollview stop, it will not properly add the new page to the beginning/end until SV comes to a rest.

If multiple ScrollViews on a page, mousewheel will only work on the first instance

Reproducible @ http://jsfiddle.net/derek/cd9Am/

Update scrollview-paginator to offset by page instead of widget container.

With the updates to scrollview-base in ticket #2532404, paginator should override host.moveTo/scrollTo and handle all movement by itself. This will enable an expected experience for a horizontal paginated carousel with each "card" being able to scroll vertically (continuous).

Weird behaviour with link selector in 3.6.0

I am migrating an app that uses the AppFramework from 3.5.1 to 3.6.0pr3 and will eventually be using 3.6.0 GA.

I understand that the AppFramework can select all the links on a page and process them using AJAX. I have been using this with 3.5.1 and it worked perfectly in IE9 and other browsers that does not support the HTML history state.

As I could not get things working with jsfiddle, I have uploaded the a simple test page to a webhost:

3.5.1: http://yuitesting.host.sk/3.5.1/
3.6.0: http://yuitesting.host.sk/3.6.0/

If using the 3.5.1 version and visiting the page in IE9, clicking the link called "test" results in this URL (with a hash fragment) being generated and no page reload: http://yuitesting.host.sk/3.5.1/#!/testtest

However, if I were to visit the 3.6.0 version in IE9, and click the link called "test", the AppFramework does not seem to intercept that click at all, and instead the browser is allowed to load that page.

Add @extensionfor API doc tags to App.Transitions and App.TransitionsNative

See Summary.

Determine which module should go into the "app" rollup

Right now the API docs for the "app" module say that everything is in there, when in fact "view-node-map" is not part of the "app" rollup. The ModelSync modules are going to be in this same boat, so I'm wondering if these extensions should be included as part of the "app" rollup.

Completed code examples omit reference to YUI library

This may seem like a minor quibble, but the code examples would not "run" on their own without the YUI library included js.
At the end of most code example pages there is a "full listing" of the source. However without the YUI "seed" nothing would work.

Perhaps this is by design, however when someone comes to this example page directly there is not even a mention of having to have a YUI seed script. It would be cool if there was a simple example of how to include the YUI seed (perhaps based off the code from the front page) which all the other code examples pages at least refer to via a simple link.

Pjax should ignore all non-HTTP URIs

Pjax currently tries to navigate on mailto: links, and it shouldn't

Investigate a default and generic Model/ModelList sync layer

Currently Model/ModelList ship with no sync layer. Any real application using the App Framework will require a sync layer, therefore that currently means the developer will have to implement one. There is a de facto Gallery module for interacting with a RESTful backend, gallery-model-sync-rest, we need to determine what we should provide out-of-the-box to help people get started quicker and make it easier for them to access their data from a remote source.

showView() should throw error if view doesn't exist

You can call showView on a view what doesn't exist - it would be great if this could throw some type of warning to aid debugging on large apps.

App.Transitions does not account for natural whitespace in the HTML

The cross-fade transition requires that all whitespace be "canceled-out" between the two views being transitioned. Natural whitespace may exist in the HTML from line-returns, spaces, tags or other characters between the HTML element tags. When this whitespace is not removed the views seem to "jump" around once the transition has finished.

The natural whitespace can be removed using the same technique that CSS Grids uses. These style rules only have to be applied while the views are transitioning, i.e. the "yui3-app-transitioning" CSS class is applied.

When a router is destroyed, all router's history event subscriptions are detached

Router needs to hold its history event subscriptions per-instance, otherwise destroying one router will detach the history event subscriptions for _all_ routers.