Custom Query

Currently ignored:

	'/dt/core/style/editor.html#test apply inline style on non-editable inline element - at non-editable inline boundary':
		'env.ie && env.version == 8',
	'/dt/core/style/editor.html#test remove inline style from non-editable inline element - at non-editable inline boundary':
		'env.ie && env.version == 8',
	'/dt/plugins/indent/indent.html#test indent next to inline non-editable':
		'env.ie && env.version == 8',

We thought that they will be fixed by #11042, but nothing has changed.

Perhaps related issues: #7895 and #8989.

There is bad TC called test avoid having inline event handlers executed in dt\core\htmldataprocessor.html
The reason is that it attempts to execute editorBot#setData multiple times (using doTest function. setData does call wait and resume inside itself so any code after first wait() call will not be executed.

Problem can be reproduced in Firefox only from CKEditor 3.6.4 rev. [7527] in both CKE 3.x and 4.x

1. Create subdomain.example.com and example.com domains
2. Put ckeditor with attached sample under subdomain.example.com
3. This sample has document.domain='example.com’ set
4. Load attached sample from subdomain (​http://subdoamin.example.com/ckeditor/samples/replacebycode2.html)
5. Press preview

Result: Permission denied error is thrown.

Error: Permission denied to access property '_cke_htmlToLoad'
Code: doc.write( window.opener._cke_htmlToLoad );
URI: ​http://www.subdomain.example.com/ckeditor/plugins/preview/preview.html

For example: ​http://ckeditor4.t/dt/core/focusManager/focus.html (#11153)

Mocked editor does not behave like the real one. It's cool to create light unit test, but most important is to write precise and realistic tests. All those which create CKEDITOR.editor instances manually are unsafe and should be rewritten.

until master

I feel that there is a case when down arrow with buttons does not work as expected. It works good with i.e. language button, color buttons - but still does not work as expected with richcombos, because it still does not focus first option.

The exception here is format menu, which checks currently active format.

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. navigate using keyboard to styles dropdown (focus it)
3. press down arrow

Expected result:
First entry in dropdown should be focused, in this case Italic Title.

Current result:
Only menu is focused.

additional info:

1. funny thing is that we don't block down arrow in this case, if scrollbar is visible (in document contianing editor), and its moveable towards bottom, then pressing down arrow will move screen

origin issue:
​t/11075

master

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. use following source code:

   <p>&nbsp;</p>
   
   <p><textarea name="foo"></textarea></p>
   
   <p><textarea name="foo"></textarea></p>
   
   <p>&nbsp;</p>
   

3. go back to wysiwyg mode, and put cursor in first line
4. press down arrow

Important note:
With firefox even more weird things happen, we are able to place cursor inside textarea, and some exception like:

Error: Permission denied to access property 'nodeType' @ http://presets.ckeditor.dev/4.3.1/full-all/ckeditor/ckeditor.js:353

Are thrown, however it has already been that way at least since 4.3 (#11262).

Expected result:
Caret should move to paragraph with textarea

Current result:
Caret skips paragraph with textarea, and moves to last line

additional info:

1. confirmed in chrome @ win8
2. it was even reproducable when i placed 2 textareas in paragraph
3. my assumption is that it's related to #11042 but it's only my gutts tell

Case 1: An error is thrown when clicking some tag in elementspath (see: screencast​).

Case 2: Confusing selection after clicking a tag in elementspath (see: screencast​).

Quite likely both cases have the same root: to be investigated.

Use this HTML:

<table border="1" cellpadding="1" cellspacing="1" style="width:500px">
	<tbody>
		<tr>
			<td>
			<p>&nbsp;</p>
			</td>
			<td>&nbsp;</td>
		</tr>
	</tbody>
</table>

1. Place caret in first table cell.
2. Move your mouse cursor to a place which will trigger magicline.
3. Press delete.
4. Error is thrown.

Uncaught TypeError: Cannot read property 'type' of null range.js:1526
CKEDITOR.dom.range.setStart range.js:1526
CKEDITOR.dom.range.setStartAfter range.js:1573
CKEDITOR.dom.range.setStartAt range.js:1618
CKEDITOR.dom.range.moveToPosition range.js:1489
CKEDITOR.dom.range.moveToElementEditablePosition range.js:2047
CKEDITOR.dom.range.moveToElementEditEnd range.js:2157
(anonymous function)

Starts in: ​https://github.com/ckeditor/ckeditor-dev/blob/c240965c7134620f8d2fe2c36f865197dc280aa9/core/editable.js#L653

Reproduced it on every browser.

1. Open replacebycode sample.
2. Click somewhere in the text.
3. Open find dialog and type "the", click "find" and close dialog.
4. Repeat 2&3 until error is thrown.

I couldn't reproduce it on Chrome and I haven't checked other browsers.

Edit:

Click somewhere in the text.

The main problem I had with reproducing this was that you have to click on "the" word like in the screen cast and not just anywhere in the text.

1. Open Opera or Chrome.
2. Go to replacebyclass.html sample.
3. Paste in the sorce mode:

   <p>lorem</p>
   <hr />
   <p>ipsum</p>
   

4. Switch to the WYSIWYG mode.
5. Put cursor after "lorem".
6. Press right arrow button.
7. Press enter.

Result: Two paragraphs have been created instead of one.

1. Open any browser (tested with IE8 and Chrome).
2. Go to "New Image plugin" sample.
3. Insert flash with dimensions (ex. 500 width; 500 height).
4. Select placeholder cut it and paste somewhere.

Result: placeholder changes size.

Moreover if you do not set dimensions of flash after cut & paste you will get width = 1 and height = 1.

Also in "Replace by class" sample everything is ok.

In IE8:

1. Open "New Image plugin" sample.
2. Click next to widget handler.

Result: you can drag widget using IE native D&D.

Checked on IE8 and IE9:

1. Open replacebycode sample.
2. Click in various places outside body, on the left side of the editor area.
3. See that selection in elements path does not reflect the place where caret is placed. Also, if you apply a style, it will be applied in different place (previous selection position).

This means that selectionChange wasn't correctly fired perhaps due to lack of native selectionchange.

Panels are too narrow, so even our default settings look bad if font is missing (FF@Fedora):

Or if someone wants a longer name:

A better solution would be to allow configure dropdowns width or, better, if editor chooses width automatically based on the content.

Based on: #11315.

There are couple of scriptLoader.load usage and none check if request completed. When request failed error should be thrown, so developer knows what happened.

For certain block elements inside CKEditor, if an element is at the beginning of the content, it is impossible to delete it using the Backspace key.

Sample elements: <form>, <blockquote>, <div>

Note: deleting does work in one particular case: only if there is no content after the element, so make sure to add some text to the content as well.

Test Case 1

1. ​http://ckeditor.com/latest/samples/replacebyclass.html
2. Paste in source mode

<blockquote>
<p>&nbsp;</p>
</blockquote>

<p>Text</p>

3. Switch to wysiwyg, place cursor in the Blockqoute element, press Backspace
4. Result: Blockqoute is not deleted.

Note 2: Pressing enter deletes Blockqoute (?)

Test Case 2

Repeat steps 1-4 with:

<form action="a" name="n">&nbsp;</form>

<p>Text</p>

(Enter key works as expected)

Test Case 3

Repeat steps 1-4 with

<div class="class">
<p>&nbsp;</p>
</div>

<p>Text</p>

(Enter key works as expected)

It is technically possible to workaround this bug by using magic line to insert first an element above (or use the Source mode), but I don't think that's a real solution for this bug.

observed at master

FF does not display images in preview.

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. click on preview button

Expected result:
You should se our sample image at right hand side.

Current result:
Image is missing, and alt is displayed.

additional info:

1. It's due to missing base tag in head. Since we're opening preview in /plugins/preview/iepreview.html, browser (correctly) assume that all relative links should be resolved starting from preview directory. So assets/foo.png will be resolved as /plugins/preview/asstes/foo.png.

Reported by a customer.

When using CKEDITOR.inline() to replace a textarea element, the width configuration setting does not work, the editor always takes the whole available width which can destroy the layout.

While investigating I've noticed that setting height do not work as well. This is in quite confusing and not documented (e.g. ​here).

Proposed solutions:

1. For width, apply the width inline style to the "main" <div class="cke_textarea_inline"> element

2. config.height should be set with min-height, to allow expanding the size of the editor while the content is added.

After call range.enlarge( CKEDITOR.ENLARGE_ELEMENT ); in some cases <\p> tag is inside enlarged range (dt/core/dom/enlarge.html 'test space 9', currently in ​git:t/10778)

<p>foo <i><b>[bar]</b></i> </p> ->
<p>foo [<i><b>bar</b></i> </p>]

and in some it is not (dt/core/dom/enlarge.html test_enlarge_element5):

<p>Test <b><i>[Enlarge</i> this]</b></p> ->
<p>Test [<b><i>Enlarge</i> this</b>]</p>

Reported through support channel.

Not sure if related to #11219.

When config.forcePasteAsPlainText is set, dragging&dropping inline widget results in widget being lost. Either drag&drop should be automatically disabled when config.forcePasteAsPlainText is set or widgets should not be lost in this very specific case.

A sample with inline widget is attached, drag & drop the widget and see that it is lost.

Tests only feature request
Since 8abc140af4644f165b9c7d227919b0a2cdfdb8fd

Since PJ created pretty convenient function to test all 3 editor modes, which was commited by PK in 8abc140af4644f165b9c7d227919b0a2cdfdb8fd. We should update our old tests which do that with custom functions (each time), since it will also simplify it.

Other candidates for such refact

ckeditor-tests-v4/dt/core/editable/events.html
ckeditor-tests-v4/dt/core/editable/readonly.html

Set allowed content to true and then insert an image map in the content:

<img alt="" src="assets/sample.jpg" usemap="#imgmap201414175321zz" />
<map id="imgmap201414175321zz" name="imgmap201414175321zz"><area coords="5,5,155,195" href="page.html" shape="rect" /></map>

Now right click on the map with the developer tools open and the tab will crash in Chrome. Adding a check in the tabletools plugin seems to be enough:

From

editor.contextMenu.addListener( function( element, selection, path ) {
	var cell = path.contains( { 'td': 1, 'th': 1 }, 1 );
	if ( cell && !cell.isReadOnly() ) {

To

editor.contextMenu.addListener( function( element, selection, path ) {
	var cell = path && path.contains( { 'td': 1, 'th': 1 }, 1 );
	if ( cell && !cell.isReadOnly() ) {

Despite this patch, I've already filed a ticket in Chrome, but this patch will avoid getting the error. ​https://code.google.com/p/chromium/issues/detail?id=331664

Related issue: #11308.

After call range.enlange there should be both opening and closing tags in range or none of them, but tests created for #10778 shown some incorrect situations:

	// <div>x<p>[foo] <b></b></p>x</div> ->
	// <div>x[<p>foo] <b></b></p>x</div>

	//Should be:
	// <div>x<p>[foo] <b></b></p>x</div>

Or:

	// <div>x<p>foo <i><b>[bar]</b></i> </p>x</div> ->
	// <div>x<p>foo [<i><b>bar</b></i> </p>]x</div>

	//Should be:
	// <div>x<p>foo [<i><b>bar</b></i> ]</p>x</div>

Note that changes introduced in #10778 does not cause this behavior, the bug existed before. To see more incorrect situations go to enlarge.html tests.

This is the continuation of #10146 issue.

Problems can be reproduced in all browsers and don't occur in CKEditor 3.x

Examples:

Two brs are handled the same way as one br

Both code snippets

<br>
<br>
<p>This is a paragraph of text.</p>

and

<br>
<p>This is a paragraph of text.</p>

will result into:

 
<p>This is a paragraph of text.</p>

This

    <br /> 
    <br /> 
    <p>This is a paragraph of text.</p> 
    <br /> 
    <br /> 
    <p>Second paragraph of text.</p>

results in

 
<p>This is a paragraph of text.</p>
&nbsp;

<p>Second paragraph of text.</p>

One br is chnaged into  

This

<table>
	<tbody>
		<tr>
			<td>Table cell contents</td>
		</tr>
	</tbody>
</table>
<br>
<p>P contents</p>

will result in

<table>
	<tbody>
		<tr>
			<td>Table cell contents</td>
		</tr>
	</tbody>
</table>
&nbsp;
<p>P contents</p>

I understand that fix for #10146 has introduced fix which changes last BR into &nbsp and most likely example two is a "won't fix". Another reason for this may be that nbsp; in second example creates in fact new line.
The only problem with example two I see is that code is in fact different to what user has entered and this doesn't occur in CKE 3.x.

Example one is rather a bug and there should be some difference between how one and two BRs are handled.

EDIT: This ticket is the follow up to #10146. Most scenarios from #10146 were fixed. Examples mentioned in this ticket are still live. When fixing this issue, please have older test cases from #10146 in mind so that they didn't get broken again.

There is a workaround mentioned in ​http://dev.ckeditor.com/ticket/10146#comment:34. It actually handles all the cases but one. When there is no other way to fix it perhaps some smarter way of using that hack could be implemented?

At the moment there's no way to specify advanced upcasting/match because enterMode in unknown for those methods (as well as other options).

Originated from #11283.

1. Open image2 sample.
2. Focus image.
3. Press ordered list button.
4. Press unordered list button.

Image's structure explodes.

<:figure class="caption cke_widget_element" data-widget="image" data-cke-widget-keep-attr="0" data-cke-widget-data='{"hasCaption":true,"src":"assets/image1.jpg","alt":"Saturn V","width":"200","height":"","lock":true,"align":"right"}'><SPAN class=cke_image_resizer_wrapper><IMG alt="Saturn V" src="assets/image1.jpg" width=200 data-cke-saved-src="assets/image1.jpg"><SPAN class="cke_image_resizer cke_image_resizer_left" title="Click and drag to resize" data-cke-expando="138">​</SPAN></SPAN><:figcaption class=cke_widget_editable contentEditable=true data-cke-expando="false" data-cke-display-name="caption" data-cke-filter="107" data-cke-widget-editable="caption" data-cke-enter-mode="2">Roll out of Saturn V on launch pad</:figcaption></:figure>

FF's implementation of focus/blur handling on nested editables seems to be very fragile - tests in ​http://ckeditor4.t/dt/plugins/widget/nestededitables.html randomly fail from time to time depending on how they were ran. Additionally, I had to add one of the tests to regressions, because it started to fail after b671945e@tests.

None of these instabilities occur when testing manually.

This is more reminder to create tests for issue #11331

since: 4.0 (didn't check earlier) until master

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. focus editor
3. press alt+f10 to focus toolbar
4. navigate to preview button
5. press button using space

Expected result:
Preview should appear

Current result:
firefofx: Preview is blocked by popup block
ie11: Exception is being thrown SCRIPT5007: Unable to set property 'location' of undefined or null reference, and popup blocker appears. Error is reported in #11597.

additional info:

1. preview is not blocked when we do this using mouse
2. tested with FF 26.0 at Win7, IE11 at Win8.1

edit: it appears that popup blocker also triggers in IE10 (@Win8), IE9 (@Win7)

since: 4.2 until master

1. open any sample with CKEditor (i.e. samples/jquery.html)
2. click maximize button in framed editor
3. click it once again to minimize editor
4. you can now edit everything in whole page

Expected result:
Body should remain untouched.

Current result:
Body element (in main window) gains contenteditable=true.

additional info:

1. I assume it's not a FF bug, coz when i checked IE11 + 4.2 (back then ie11 was recognized as gecko) it shown exaclty the same deffect

1. Open editor (replacebyclass.html) and paste following html as a source:

   <ol>
   	<li>aa
   	<blockquote>
   	<ol>
   		<li>bb</li>
   		<li>cc</li>
   	</ol>
   	</blockquote>
   	</li>
   	<li>dd</li>
   </ol>
   

2. Go to WYSIWYG mode and select [bb ... cc].
3. Press "Insert/Remove Bulleted List" button.

Result:

• nested list type is unchange,
• error in console: "Uncaught TypeError: Cannot read property 'parent' of undefined."
• when you do to the source mode you will se bookmarks:

  <li><span id="cke_bm_152S" style="display:none">&nbsp;</span>bb</li>
  <li>cc<span id="cke_bm_152E" style="display:none">&nbsp;</span></li>
  

Tested with Chrome and FF.

1. Open any page with CKEditor in Safari.
2. Maximize browser window (an arrow in the upper right-hand corner).
3. Open a dialog (link, image).
4. ESC.

Expected: Dialog is closed. Actual: Dialog is closed but window is no longer fullscreen.

My first guess is that some preventDefault is needed. It is not reproducible in Chrome though, so additional research is needed to figure out if the problem is CKEditor or Safari itself.

1. Open plugins/mathjax/samples/mathjax.html
2. Wait for formulas to load.
3. Double click a formula, don't change or touch anything.
4. Press ESC.

Expected: Dialog should close

Actual: Editor says that some dialog contents has been altered (but really hasn't).

5. Press ESC to get rid of the window.

Expected: Dialog should remain open (even though the prompt is invalid), cancel should prevent dialog from closing.

Actual: Sometimes another prompt is displayed which, when ESCed, sends us back to the editor. Sometimes no additional prompt is displayed but this is still invalid because dialog is closed.

I'm unable to reproduce in other browsers/dialogs.

I managed to reproduce it with sourcedialog, randomly (see another screencast).

1. Open placeholder plugin.
2. Select "is a sample placeholder. You".
3. Drag it and drop at the end of paragraph.
4. Drag&Drop widget using drag handler.

Result: It is not possible to d&d widget.

Tested on Chrome, FF and IE 11. On each I get different results but on any of them I was not able to drop widget.

1. Go to placeholder plugin.
2. Select whole content of the editor.
3. Copy (Ctrl+C).
4. Paste into notepad (or similar).

Result on Chrome:

This is a [[sample placeholder]]Click and drag to move. You are using CKEditor.

On FF:

This is a [[sample placeholder]] [Click and drag to move] . You are using CKEditor.

"Click and drag to move." should not be there.

1. Open IE8 and go to replacebyclass.html.
2. Put cursor after Apollo 11 header.
3. Press "Replace" button.
4. Set Find what: Apollo and leave Replace with: empty.
5. Double click on Replace button so the second Apollo (beginning of the first paragraph) will be replaced by empty string.
6. Press "Cancel" button.

Result: You cannot close the dialog and error in the console: 'type' is null or not an object.

Since 4.0.

I've tested this with Chrome and IE 10 and everything is fine there.

1. Open placeholder.html sample in IE 8.
2. Copy sample placeholder and paste it at the end of paragraph.
3. Drag the second placeholder (using drag handler) and drop in directly after the first one.

Result: Placeholder disappears.

Sometimes it is possible to drop it correctly.

Sometimes I get "Unspecific error" doing the same with image inline widgets.

1. Open image2 plugin in IE 8.
2. Click Image button and open "Image properties" dialog.
3. Set http://upload.wikimedia.org/wikipedia/commons/6/63/Wikipedia-logo.png as a URL (wikipedia logo with width 200 and height 200).
4. Change the width of the image to 20.

Result: height is 21 instead of 20.

5. Change the width of the image back to 200.

Result: height is 214 instead of 200.

Clearly IE 8 do not know how to math.

In Chrome everything is ok.

In 4.3-beta everything was fine.

1. Open replacebyclass.html sample in IE 8.
2. Select content: from strike through three and up to the begging of the document.
3. Press Ctrl+C or Delete.

Result: Error: Invalid argument and Browser crash.

The same in inline.

IE9 and Chrome works fine.

Since 4.0.

Buttons in elementspath should not be draggable. Now you can drag it into editable. What's worse in IE10, IE11 it will cause following exception:

SCRIPT5009: 'CKEDITOR' is undefined 
replacebyclass.html, line 355 character 1

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. click somwehere inside editable
3. click move body elements path part into editable

additional info:

1. it should be easily fixable with setting onmousedown="return false" on anchors

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. use "Find" button from toolbar
3. (you'll have Find what input focused) start pressing space

Expected result:
Spaces should be placed at the beginning of input

Current result:
No spaces are placed (they may be placed at the end of the input though)

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. clear editor contents
3. insert new image
4. set following properties

• URL: ​http://i.imgur.com/Gj8iBon.png
• alternative text: altText
• border: 2
• hspace: 2
• vspace: 2

5. double click at image (in order to open properties dialo)

Expected result:
Image properties dialog should be shown.

Current result:
Browser crashes.

additional info:

1. sometimes you're able to repro it without setting alt text

TC 1 (general, IE9)

1. Open samples/replacebyclass.html
2. Insert page break.
3. Select inserted page break.
4. C-c
5. Change selection.
6. C-v

SCRIPT5007: Unable to get value of the property 'isBlock': object is null or undefined 
editable.js, line 1708 character 5

Works in 3.x.

TC 2 (IE11)

1. open any sample with CKEditor table plugin (i.e. samples/replacebyclass.html)
2. switch to source mode, and use following source code:

   <div style="page-break-after: always;"><span style="display:none">&nbsp;</span></div>
   <table border="1" cellpadding="1" cellspacing="1" style="width:500px">
   	<tbody>
   		<tr>
   			<td>
   			<p>aa</p>
   
   			<p>bb</p>
   			</td>
   			<td>&nbsp;</td>
   		</tr>
   		<tr>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   		</tr>
   	</tbody>
   </table>
   

3. switch back to wysiwyg mode
4. copy page break into first (top left) table cell

Expected result:
Operation should not cause exception.

Current result:
Exception is being thrown:

SCRIPT5007: Unable to get property 'isBlock' of undefined or null reference
File: ckeditor.js, Line: 323, Column: 45

How to reproduce.

1. Go to the demo Page ​http://ckeditor.com/demo#full
2. Switch to source mode
3. Delete all the content
4. enter the following content

   <iframe src="http://google.de"><!-- Hello -->This is a test</iframe>
   

5. switch to wysiwyg mode
6. switch back to source mode enter the comment is messed up

<iframe src="http://google.de">&lt;!--{cke_protected}{C}%3C!%2D%2D%20Hello%20%2D%2D%3E--&gt;This is a test</iframe>

There are several things concerning the dialog that should be investigated/fixed/re-factorized:

1. Keyboard button labels are not consistent

   Editor Context Menu Press Shift+Ctrl+F10 or APPLICATION KEY to open context-menu. Then move to next menu option with TAB or DOWN ARROW. Move to previous option with SHIFT+TAB or UP ARROW. Press SPACE or ENTER to select the menu option. [...]

This is because some labels are hard-coded (upper case) and some are generated from langfiles (lower case). #11196 allowed translation of keys and we should take advantage of that feature. All hard-coded labels must be converted into dynamic ones.

2. We must verify whether help strings still match editor features. Some of them may be outdated/irrelevant/misleading. If necessary, some new ones should be written.

3. Some instructions are too complicated, some are unclear, e.g. lots of translators do not understand the MagicLine-related texts so we should revise the contents overall.

4. We must investigate a11y aspect of the dialog, e.g. compatibility with screen readers (JAWS). Perhaps <kbd> tags would be suitable here and there.

Related tickets we could also take into consideration:

• #8490: Plugins should extend a11yhelp to make the dialog dynamic and (build/editor)-dependent. At the moment a11yhelp creates a delusion that some features are enabled while, in fact, the plugins to which they belong are not even loaded.
• #9638: "Press ALT 0 for help" iframe voice label makes no sense when a11yhelp is not loaded (i.e. basic build).
• #10929: Spelling Mistakes on Accessibility Instructions dialog.

env.ie6Compat ... env.ie9Compat are deprecated since 4.0. After clean-up (#11422) these variables are rarely used:

env.ie9Compat -> 1 time
env.ie8Compat -> 2 times
env.ie7Compat -> 0 times
env.ie6Compat -> 0 times

They should be finally removed.

The problem with quirks appeared after removing IE 6 & IE 7 support (#11422), because most of hacks now apply only to QM (instead of IE 7, IE 6 and IE QM). Even worse it is implemented in multipile ways (unsing "env.ie && env.quirks", "env.ie6Compat" or "document.documentMode").

I think that it would be misleading if we keep if ( env.ie6Compat ) in our code only because of QM (env.ie6Compat is true in QM) and it is better to replace it with if ( env.ieQuirks ).

Also after clean up we can save some bites (compressed ckeditor is 322 byts smaller with env.ieQuirks).

I'm not sure if there is a point in keeping env.quirks == true for non-IE browsers. Maybe env.quirks should be true only in IE 8 and IE 9 QM (so we will use just env.quirks instead of env.ieQuirks).

See e.g. #11295. Calling editor.destroy() right after (or before it's completed) CKEDITOR.replace() or editor.setData() causes errors.

That's because developer has to care about calling methods when previous finished. There are callbacks and events which notify developer when action is completed, but it's his duty to find which method can be called when.

However, not every developer understand problems with asynchronous methods and not in every case it's easy to handle this. For example when editor is loaded in one tab of some UI component and user switches between tabs too quickly, destroy() may be called before editor fully initialises.

On the other hand, if editor's methods would take care of that (e.g. destroy() would wait until initialisation finished), then API would start to work unpredictably. Developer would never know if destroy() will be done immediately or if it's going to wait until something (setData, initialisation, etc.) ends. This may be even worse situation than the current one.

We could make even longer step and make all editor's methods asynchronous and e.g. based on promises. Then everything would be fine... if you understand all of this :D.

Therefore, instead of forcing developers to understand some not trivial concepts to do basic stuff we can simply clarify this in docs. But there still will be cases in which it will be hard to handle editor state (like the tabs case).

Opinions needed.

More details about the current situation in ​http://dev.ckeditor.com/ticket/11502#comment:8

Cases we need to solve

1. Integration with styles applicable to widgets. See #11297.
2. Disabling ACF in body of some widgets - e.g. mediaembed. See #11737 and #10925.
3. Filtering pasted widgets. See #11115.

Cause of problems

1. ACF filters data before widgets are upcasted. Therefore for example mediaembed cannot notify ACF that children of some elements should not be filtered and it's not possible to create allowed content rule based on a widget name and style properties allowed on that widget element.
2. In pasted or dropped content we have the upcasted and initialized HTML of a widget. Therefore we cannot filter it, because ACF is applied to the data format.

Solution

ACF after upcasting

The advantage of this solution is that ACF would know everything about widgets, so it could make precise decisions. Additionally, while filtering pasted content there would be no problem at all, because it would be the same case.

However, there are two problems which makes this idea incorrect:

• Assuming that we've got some complex widget like image2, which makes a lot of transformations while upcasting, how would ACF filters the upcasted image? How could it strip, without breaking widget, figcaption and figure if they are not allowed? It's simply impossible, without some special actions from the upcast method. It could theoretically ask ACF if specific components of widget are allowed, but it's tricky and still doesn't solve this issue completely.
• Moreover, when ACF would filter upcasted version of widgets it would be impossible or a lot harder to define ACF rules. One would have to know internal widget structure to define those rules. In my opinion this makes the idea totally wrong. ACF filters data format and that's a foundation we have to keep.

ACF filters data

Previous section proved that ACF needs to be applied to data, not to inner HTML. This means that the current way of processing is the only correct one, but on the other hand we still have those three cases, which are listed at the beginning, that have to be solved.

1. Integration with styles applicable to widgets - a solution has been proposed in ​http://dev.ckeditor.com/ticket/11297#comment:8. It may not be 100% precise in some cases, but there's another ACF's characteristic which we have to keep in mind. It's not meant to be a security filter. It's meant to ensure good UX in as many cases as possible. Therefore when uncommon constellation of element and its attributes slips through its hands it's not a blocker.

2. Disabling ACF without a possibility to communicate through element's attributes. In this case the only possible solution is a callback executed by ACF which will notify it about how it should proceed. For example mediaembed will add a callback which will tell ACF that all descendants of element should be skipped if element has mediaembed's class. Callback will return binary flags, because we may need more options in the future. Example of flags that we may need to implement:

• DISABLE_ON_ELEMENT - skips only this element (continues on descendants),
• DISABLE_ON_DESCENDANTS - skips descendants,
• CONTINUE_ITERATION - continue iteration - used with DISABLE_ON_DESCENDANTS could inform ACF that it should look for a descendant that reenables filtering,
• ENABLE - reenable filtering.

But of course for now we'll need just one of these flags, so only one should be implemented.

Ticket: #11737.

3. Filtering pasted widgets. It's not possible to filter widget's upcasted version - only its source version can be filtered. If we had full control over clipboard we could downcast selected data on cut/copy and then filtering on paste would work like filtering on data load. Unfortunately, we don't have full control over clipboard and won't have even after #11437, because of IEs.

Therefore, the only solution might be to look for upcasted widgets in pasted data, extract their HTML, find out from which editors they came from, use those editors to downcast them, and replace their upcasted version with downcasted one in pasted data. This could be based on unique editors' ids and data-cke-widget-editor-id attributes marking widget wrappers. This won't work when copying between window tabs or frames or when editor instance has been destroyed in the meantime and I'm not sure what should happen in such situation.

Similar solution, maybe more straightforward and more reliable would be if editor, on cut/copy/dragstart if it cannot override the content, could retrieve widgets' HTML from selection, downcast them and store in the CKEDITOR object, so all other editors will be able to replace pasted widgets with their downcasted HTML stored in that object. Of course widgets have to be marked with unique ids to make this possible. This option is better than the previous one because widget's HTML will be kept even after editor is destroyed. We can also reuse the same mechanism in drag and drop for IEs in which we will need to store the data for a while. As the mechanism will be public 3rd party code will be able to integrate with it.

And the third solution, nearly perfect, is to control copy and cut and override data with HTML extracted by us and processed by htmlDP.toDataFormat. We will need to do the same for drag&drop, so it won't be a big problem to eventually make the same with copy/cut. Additionally, we would automatically solve the problem with Blink putting awful HTML into clipboard.

Extracted from ​http://dev.ckeditor.com/ticket/11439#comment:16

Due to invalid focus in anchor dialog we can experience few issues. You're not able to:

• focus other fields by pressing tab
• close anchor dialog with esc

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. using Source button set html to:

   <p><a href="#foo">bar</a></p>
   

3. return to wysiwyg mode
4. double click link bar
5. press esc key

Expected result:
Dialog should be closed (as it is for typical links)

Current result:
It's not closed unless you'll focus any input control inside of it.

additional info:

1. Tested in Opera (blink), IE10, FF 27.0.1 @ Win8
2. Issue was also present in 4.0 release (didn't check earlier releases)

TC1:

1. Open image2 sample.
2. Doubleclick image.
3. Change the width to 300.
4. See that height was set too although it was empty at the beginning. I'd expect that it was left empty.

Notice that in step 2 "keep ratio" is on. I think that it's ok, because it tells user that ratio will be kept if he changes size, which is true because the missing dimension is calculated automatically.

TC2:

1. Open image2 sample.
2. Doubleclick image to verify that only width is set. Close the dialog.
3. Resize image using handler.
4. Doubleclick image.
5. See that both dimensions are set now.

• Edit: TC2 is not about reading dimensions by dialog but about editor calculating both dimensions. In HTML when only one dimension is set the other one is automatically calculated by browser so we should not interfere it.

See attached image.

It's a Firefox bug: ​https://bugzilla.mozilla.org/show_bug.cgi?id=965966

Extracted from #11298

Currently MathJax supports only inline equations. It would be nice if block equations were supported as well.

Additionally plugins should detect what sort of equation it is dealing (probably based on brackets used).

This is umbrella ticked for MathJax improvements.

The list of the tickets with bug/feature requests for MathJax:

• #554 - MathML support,
• #11595 - block equations support,
• #11313 - https support,
• #11298 - fix baseline.

See ie-selection-cached.webm​. Tested on IE8 and IE9.

1. Write few paragraphs.
2. Select all using mouse.
3. Press list button.
4. Click once in the second item.
5. Press tab to indent it.

Result: entire list was indented. Most likely editor.getSelection() returned cached selection made in step 2.

1. Open "UI Color Plugin" sample.
2. Press "UI Color Picker" icon.
3. Change color.
4. Close dialog using "X" in the top right corner.

Result: Alert "You have changed some options. Are you sure you want to close the dialog window?" is shown but changes are applied anyway.

Expectation: Alert will not be shown or changes will be reverted when I close dialog using "X".

I am for first option so alert should not be shown.

1. Create editor with config.fullPage = true
2. Set HTML:

   <html>
   <head>
   	<title></title>
   </head>
   <body dir="rtl"></body>
   </html>
   

3. Show source/getData.

Expected: The same HTML as set in 2.

Actual:

<html>
<head>
	<title></title>
</head>
<body dir="ltr"></body>
</html>

The origin I discovered the bug with docprops plugin, which brings Language Direction field in its dialog.

What's wrong? First bad commit is ​git:751e298cca8. It's fine in 3.6.x.

1. Use IE11.
2. Set data

   <ul>
   	<li>x</li>
   	<li>y</li>
   	<li>z</li>
   </ul>
   

3. Focus any list item.
4. Click "ul" in Elements Path.
5. Click indent: nothing happens (#11604).
6. Put selection

   <ul>
   	<li>x^</li>
   	<li>y</li>
   	<li>z</li>
   </ul>
   

7. Click indent.

Expected (considering that 5. is right):

<ul style="margin-left: 40px;">
	<li>x</li>
	<li>y</li>
	<li>z</li>
</ul>

Actual:

<ul>
	<li>x</li>
</ul>

Two list items are gone. Just like that. This ticket may be related to #11604.

1. Open "Replace By Class" sample.
2. Click after "humans" to move cursor.
3. Press shift and click after "Neil".

What is the expected behavior? ", Americans Neil" is selected.

What went wrong? Cursor moved instead of select text.

This is actually a Blink/contenteditable bug and I reported it to the chromium project: ​http://code.google.com/p/chromium/issues/detail?id=345745&thanks=345745&ts=1393005706

Chrome reports this in developer console, when I click inside wysiwygarea:

body.scrollLeft is deprecated in strict mode. Please use 'documentElement.scrollLeft' if in strict mode and 'body.scrollLeft' only if in quirks mode.

This code do not produce any issues, however I'm not sure if users aren't scared a bit of it, for example.

Default configuration:

CKEDITOR.replace( 'editor1', {
	toolbarGroups : [
		{ name: 'document',    groups: [ 'mode', 'document', 'doctools' ] },
		{ name: 'clipboard',   groups: [ 'clipboard', 'undo' ] }, // <----
		{ name: 'editing',     groups: [ 'find', 'selection', 'spellchecker' ] },
		{ name: 'forms' },
		'/',
		{ name: 'basicstyles', groups: [ 'basicstyles', 'cleanup' ] },
		{ name: 'paragraph',   groups: [ 'list', 'indent', 'blocks', 'align', 'bidi' ] },
		{ name: 'links' },
		{ name: 'insert' },
		'/',
		{ name: 'styles' },
		{ name: 'colors' },
		{ name: 'tools' },
		{ name: 'others' },
		{ name: 'about' }
	]
} );

Undo subgroup removed (undo/redo buttons are gone):

CKEDITOR.replace( 'editor1', {
	toolbarGroups : [
		{ name: 'document',    groups: [ 'mode', 'document', 'doctools' ] },
		{ name: 'clipboard',   groups: [ 'clipboard' ] }, // <----
		{ name: 'editing',     groups: [ 'find', 'selection', 'spellchecker' ] },
		{ name: 'forms' },
		'/',
		{ name: 'basicstyles', groups: [ 'basicstyles', 'cleanup' ] },
		{ name: 'paragraph',   groups: [ 'list', 'indent', 'blocks', 'align', 'bidi' ] },
		{ name: 'links' },
		{ name: 'insert' },
		'/',
		{ name: 'styles' },
		{ name: 'colors' },
		{ name: 'tools' },
		{ name: 'others' },
		{ name: 'about' }
	]
} );

Clipboard subgroup removed (clipboard buttons are still available):

CKEDITOR.replace( 'editor1', {
	toolbarGroups : [
		{ name: 'document',    groups: [ 'mode', 'document', 'doctools' ] },
		{ name: 'clipboard',   groups: [ 'undo' ] }, // <----
		{ name: 'editing',     groups: [ 'find', 'selection', 'spellchecker' ] },
		{ name: 'forms' },
		'/',
		{ name: 'basicstyles', groups: [ 'basicstyles', 'cleanup' ] },
		{ name: 'paragraph',   groups: [ 'list', 'indent', 'blocks', 'align', 'bidi' ] },
		{ name: 'links' },
		{ name: 'insert' },
		'/',
		{ name: 'styles' },
		{ name: 'colors' },
		{ name: 'tools' },
		{ name: 'others' },
		{ name: 'about' }
	]
} );

Using timestamp to mark unique builds worked fine in the past where we released packages once per ~month.

Now, when lots of users try different builds, overwriting the same location and then scratching their heads why something does not work, we need to do something different. See for example ​http://ckeditor.com/forums/CKEditor/Bug-with-custom-builder-and-language-selection, there were more issues reported like this.

My suggestion is to go with assigning a random hash when creating a build. There is no sense in calculating the hash based on the list of enabled plugins (and their versions, btw. we do not have this information in case of using command line builder), languages etc. Using a hash that is complex enough will do the trick.

Using 6-characters long hash that consists of small letters + numbers, gives ~1 838 265 625 (35^6) combinations if I counted correctly (1.8 billion).

Add to contents styles:

img { margin: 10px }

And the resizer will be displayed 10px from its correct position. Can we do anything about it other than resetting that style?

I think that the only solution is to change the resizer position if margin was discovered, but that's not even close to a clean solution.

On the other hand, if I set padding instead of margin, then resizer also isn't displayed over the image. But in this case it's still displayed over the widget outline (so it looks good), because the outline is also pushed from image. So theoretically we could add border or ~0px padding to widget wrapper, so it'd be pushed from image if it has margin, but then we would break margins collapsing between image and surrounding elements, what's not acceptable.

1. Insert below code on page and into contents.css

   var editor = CKEDITOR.replace( 'editor1', {		
   				//allowedContent : true,
   				colorButton_colors : 'highlight_yellow/FFFF00',
   				colorButton_backStyle : {
   					element : 'span',
   					attributes : {
   						'class' : '#(color)'
   					}
   				}
   });
   
   
   //in contents.css
   .highlight_yellow{
   	background-color: #FFFF00;
   }
   

2. Refresh the cache.
3. Highlight the word and change its color to Yellow.

Results:

1. If you use custom color for which translation is not defined in language file e.g. 'highlight_yellow/FFFF00' then you will see green square in color button dropdown panel but its label is 'FFFF00'. Maybe I'm wrong here but I would expect label to be displayed as 'highlight_yellow'.

2. The 'highlight_yellow' is used for styling text you select but it has no effect. Such style <span style="color:highlight_yellow">item 2</span> is not recognized thus it looks like text is not styled.

3. With ACF enabled, once you switch to source and back, span will be removed and it shouldn't be. This part is reproducible from CKEditor 4.1.

4. You can't change this color to automatic with such configuration. This part can be reproduced from CKEditor 3.0.

Clicking text in editable moves caret to very beginning, which is extremely annoying if you want i.e. select something in order to bold it.

1. open sample with image2 plugin (/samples/plugins/image2/image2.html)
2. put caret at following position Roll out of |Saturn V on launch pad in first widget caption
   note first selection is OK
3. move caret to other element (by clicking), i.e. "Apollo 11" headline
4. repeat 2nd step

Expected result:
Caret should be placed in of |Saturn

Current result:
Caret is placed at very beginning of the caption.

additional info:

1. Bug occurs only if focus is moved from within the editor. That means, if you'll blur editor by i.e. clicking at sample headline "CKEditor Samples » New Image plugin", and then you'll click at editable - bug does not occur.

1. Open placeholder sample (plugins/placeholder/samples/placeholder.html).
2. In the source mode enter:

   <p>This is a [[sample ]placeholder]].</p>
   

3. Switch back to the WYSIWYG mode.

Result: there is no placeholder.

4. In the source mode enter:

   <p>This is a [[sample >placeholder]].</p>
   

5. Switch back to the WYSIWYG mode.

Result: there is a placeholder.

6. Open placeholder dialog.
7. Press "Ok".

Result: Error message:

The placeholder can not be empty and can not contain any of following characters: [, ], <, >

On the one hand it is not possible to upcast placeholder with ']' character, because it is forbidden, but on the other we can do it with '>'.

In my opinion it should not be possible to upcast placeholder with forbidden character but also it would be good if user would be informed that the placeholder contains forbidden character.

checked only on IE8, but this issue might be also present in other IE versions

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. using "Source" button set content to following markup:

   <p>aa bb<u> cc </u>dd ee</p>
   

3. go back to wysiwyg mode
4. place caret in following position aa bb cc |dd ee
5. press ctrl + shift + right arrow

Expected result:
Selection should be expanded to contain dd string.

Current result:
Selection is not expanded.

additional info:

1. i've tested it with bold, italic - they have the same issue too

1. Open Replace by Class sample.
2. Open Link dialog.
3. Paste long link into URL field (longer than text field width) like:

   ckeditor.com#Lorem-ipsum-dolor-sit-amet-consectetur-adipiscing-elit-Nam-vel-lectus-varius
   

4. press "Home" button

Result: there is a space between cursor and text.

The same issue for 'end' button. I was able to reproduce it on IE9 and IE10 with all of the text fields. On IE8, IE11 and Chrome everything is fine.

We should think about providing good a11y for widget.

Currently screen readers treats every widget as the end of an element. We need to do far better than that.

The most important requirements i see at the moment are:

• labels for widgets
• labels for editables

labels for widgets

• Should be context sensitive, by that i mean they should be diffrent for image, placeholder, mathjax and codesnippet
• Label should be also possible to fetch some information from widget instance itself, i.e. image should be have its alt attribute included into label
• Should be specified in widget definition

labels for editables

Here i have no clear conception as of yet, because you're only able to access editable using the tab key, but it iterates from the very beginning of the document, rather than current caret position.

Currently 2 solutions come to my mind:

Solution 1

• leave current keyboard as it is
• upon getting into editable, user should be informed what widget editable belongs to

Solution 2

• Allow only to enter into editable (with tab key) only when widget is focused
• We don't need to inform our end-user what widget he's in (that reduces extra time spent on listening). The only one information he will need is the name of editable itself
• Important implementation detail would be to allow focus cycling inside widget

Insert below code into replacebycode sapmle:

editor.on('instanceReady', function(){
	var elem = new CKEDITOR.dom.element( 'pre' );
	editor.insertElement(elem); //error in IE
	//editor.editable().append(elem); //works
});

When you load the page you will get:
Message: 'undefined' is empty or not an object
Line: 299
URI: /ckeditor4_git/core/editable.js

Problem can be reproduced from CKEditor 4.3 in all versions of IE.

Tested on the Safari (537.51.2) on iOS 7.1 (iPad).

Dialogs are in the correct position as long as I'm not using zoom.

Tested on CKEditor 4.3.4, Chrome 33 on Android 4.4.2.

Android change font size in the contenteditable area what is on the one hand useful, because it is easier to read and edit content. But on the other hand, because of this, non-editable elements, like placeholder widget, has smaller font size. Also what user see is not what he would get.

We should search for a flag to disable such feature and consider if we should use it.

Tested with Safari (537.51.2) on iOS 7.1 (iPad).

Magicline does not work on Safari on iOS.

It could works as it works on Chrome on Android so show the magicline when cursor is just before or after the position of the magicline. Or magicline could be shown when user tap on the magicline position. There are solutions.

Tested with Safari (537.51.2) on iOS 7.1 (iPad), CKEditor 4.3.4.

When I apply any command page scoll down so toolbar is above the viewport.

Tested with Safari (537.51.2) on iOS 7.1 (iPad), CKEditor 4.3.4.

When I try to insert a content using dialog (link, special character, smiley) cursor is moved to the begging of the document.

Tested with Safari (537.51.2) on iOS 7.1 (iPad), CKEditor 4.3.4.

1. Open "Replace by Class" sample.
2. Press "Source" button.
3. Press "Source" button again.

Expected: editor will switch back to the wysiwyg mode.

Result: nothing happens.

I can go back to the wysiwyg mode if I move the focus to the source textarea.

Tested with Safari (537.51.2) on iOS 7.1 (iPad), CKEditor 4.3.4.

1. One Image2 sample.
2. Tap on the image.
3. Press image button in the toolbar.

Expected: Image edit dialog will be shown.

Result: Empty dialog is shown.

Tested with Safari (537.51.2) on iOS 7.1 (iPad), CKEditor 4.3.4.

Using build-in format tools I'm able to execute commands which are forbidden by ACF (see attachment). Also modification made by this options are on recorded by undo manager.

Follow-up of #11480.

We should add some support for tab key inside code-edit textarea of codesnippet dialog. We need to handle following keys:

1. Tab - should insert a tabulation character
2. ctrl + . (period) - should move focus to next focusable item in dialog

Note: that shift + tab should remain untouched.

We're considering 2 solutions:

1. Handling it completly by ourself (leftovers are in ​t/11480b branch)
2. Use external library, like ​Behave.js
3. Or use full code editor.

Follow up of ​http://dev.ckeditor.com/ticket/8587#comment:8

Using position:fixed for the editor chrome may be beneficial for us because it will need less code, there won't be a problem with input names, it may be faster and more stable. Additionally, we may check if there are other ways to remove scrolls from the viewport which will not cause so much damage underneath the editor.

Problem can be reproduced in IE8-10 (works in IE11).

1. Add attached insertVariable plugin (best in CKEditor dev)
2. Add below code in HTML page:

   CKEDITOR.replace( 'editor1', {
   	extraPlugins: 'insertVariable'
   });
   

3. Load the page and open insertVariable Dialog.

Result: in IE8-10 there is always vertical scrollbar. This is happening because below TD element exceeds size of wrapper div:

<td class="cke_dialog_ui_vbox_child" role="presentation" style="width: 100%; height: 100%;">

<div class="cke_dialog_ui_vbox cke_dialog_page_contents" id="cke_111_uiElement" role="tabpanel" aria-hidden="false" aria-labelledby="cke_iframe_112" style="width: 100%; height: 100%;" name="iframe">

This scrolbarr is not a big issue with small content but you you use large contant you will get double vertical scrollbar (Please see attached iframeDialog.png)

One solution to this problem is adding below rule in ckeditor/skins/moono/dialog_ie.css:

.cke_dialog_page_contents {
	overflow:hidden;
}

The above rule solves the problem (please note that kama skin uses this class in dialog_iequirks.css) but if someone uses ony iframe in his dialog it would be nice idea to to have paddings removed on cke_dialog_contents_body. Perhaps we could introduce separate class for iframeDialog and 'contents body' or simply modify both?

1. Editor styles.js so Styled image names are shorter (to make them visible in the button.
2. Load <p><img alt="" src="assets/sample.jpg" /></p>.
3. Select image and apply one of the styles.
4. See that the styles dropdown still shows "Styles" instead of the applied style name.
5. Deselect and select image again - styles dropdown shows the applied style name.

Reproduced on master and major. Checked Firefox and Chrome.

I stumbled upon this issue while developing sample for #11480.

This issue makes CKEditor events API quite useless without editor instance, especially if the editor is destroyed in the callback.

See ​jsFiddle.

How to reproduce?

1. Create editor instance.
2. Attach event listener on a DOM element outside of the editor. Not related to editable, etc.
3. See that event works.
4. Destroy the editor(s).

Actual:

5. Attached listener is not fired anymore.

Expected:

5. Attached listener is fired since the object in DOM is still there.

Quick research:

• ​https://github.com/ckeditor/ckeditor-dev/blob/master/core/dom/domobject.js#L150-L152
• #5663

Div is not replaced with editor when you'll click inbetween paragraphs. It's minor issue but can mess up user UX, if he'll click in such place for the very first time.

1. open divreplace sample (​divreplace)
2. click on space between two paragraphs

Expected result:
Editor should be created.

Current result:
Nothing happens.

additional info:

At the moment (#11297) to change the way widgets are styled, developer must override widget#add/has/getClasses, which is not quite straightforward.

The idea is to introduce widget#styleableElement property which, referring to the right element, can be re-used in widget#add/has/getClass and reduce the complexity of customisation.

Problems:

• How to set it? Via widgetDef (selector, method...)? Via widget.setStyleableElement? By overriding widget.getStyleableElement?
• How to alter it? delete widget.styleableElement? widget.setStyleableElement? By overriding widget.getStyleableElement?
• How to optimise it? Cache? How to override cached element?

1. Open plugins/image2/dev/image2.html
2. Focus captioned image. In framed editor format shows that none of the formats is used, in inline editor format shows "Normal (DIV)".

Plugin should allow to use other highlight.js path than the default one. It will allow developer to:

• link it against more convenient path, where it's already used for website front-end - thus he might update highlight.js in one place
• link it to other cdn

Create a "test.xml" file with something like this:

<?xml version="1.0" encoding="utf-8" ?><Templates><Template>content</Template></Templates>

Add this to a page with CKEditor:

CKEDITOR.on('instanceReady', function(e) {
	CKEDITOR.ajax.loadXml( "test.xml", function(oXml) {
		var child = oXml.selectSingleNode( 'Templates' );
		if (!child)
			alert("Failed, the Templates node hasn't been found");
		else
			alert("XML successful");

	});
});

Now when the page is loaded IE11 will state that the child hasn't been found This can be prevented by using the XML code found in CKFinder (in theory it was added for Android, but it turns out that it also works here).

As in a ticket title.

It seems that we have couple of issues that concern Asian languages and input methods:

1. #8854, #10414, #13575 - IE, Enter_BR result in invalid input. Related ticket - #12391
2. #12597 - Blink, Enter_BR result in invalid imput
3. #9332, #9749 - startupFocus:true in Firefox, not possible to type Japanese chars
4. #11663, #12415 - IE11, autogrow result in invalid input
5. #11601 - Firfox space is increased on copy/paste
6. #8913 - full width spaces.
7. #11764 - not yet confirmed.

8. #12910 - IE11, Korean; Issue may be associated with string size.
9. #13043 - IE11, Korean; Inline styles are not maintained when typing.

Click on the left or right margin of editable - caret won't be moved to the closest possible solution. This is very bad for UX, because it's hard to place caret at the beginning of paragraph.

The solutions should be easy - use padding-left/right instead of margin-left/right which causes that body does not start from left:0 and does not end at right:100%, but is padded from viewport border.

Using paddings will break margins collapsing, so to avoid breaking more often used margin-top/bottom (for paragraph, headers, etc.) we should still use margin-top/bottom for the body.

If there's a different way, like using styling for HTML element, which could perhaps keep left/right margins collapsing too, I'd gladly see this solution.

1. Open replace by class sample.
2. Click on "Apollo" image.
3. Open image dialog.

Image is align to right, but "Alignment" is "<not set>". It is because we use class instead of inline style to align this image.

1. open any sample with CKEditor (i.e. samples/replacebyclass.html)
2. use following source code

   <table border="1" cellpadding="1" cellspacing="1" style="width:500px">
   	<tbody>
   		<tr>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   		</tr>
   		<tr>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   		</tr>
   		<tr>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   			<td>&nbsp;</td>
   		</tr>
   	</tbody>
   </table>
   
   <p>&nbsp;</p>
   

3. switch back to wysiwyg mode
4. place cursor in very last cell (9th)
5. press ctrl + backspace twice

Expected result:
Caret should not move outside of the cell?

Current result:
Leaves table with one column and two rows

additional info:

1. it acts weird with ctrl+backspace in many other cases with tables

Following #11341, it's not possible to create an anchor (with Anchor button) out of image2 widget (image) or, at least, the feature is buggy in most cases. Since it's possible to create linked images, users would expect to do the same with anchors.

There are two solutions: We can either completely disable the feature but, quite frankly, it would not make much sense since linking already works or we can simply enable it.

Special case of #11963.

1. Open editor with no content
2. Focus editor
3. Click few times "Increase indent"
4. Click "Insert/Remove ordered list"
5. Switch to source mode:

<ul>
	<li style="margin-left: 120px;">Foo</li>
	<li>Bar</li>
</ul>

Expected:

<ul style="margin-left: 120px;">
	<li>Foo</li>
	<li>Bar</li>
</ul>

Reasoning

Paragraph's margin should be moved to ol/li when creating list, because this is where indent/outdent buttons add/remove margin when you have a caret in the first item of a list.

However, I'm not sure what if we're creating list out of few paragraphs when each have different margin. I think that in such case it's best to remove those margins and "normalize" the situation. Otherwise, we'd have to go crazy and e.g. create sublists based on indentation of following paragraphs, but that would be a waste of time and we don't know if user wanted to do that anyway.

Alternatively, we can simply remove all margins when creating list, because none of the solutions seem to be 100% correct. Everyone can have different idea about how that should work.

1. Open placeholder sample ​http://presets.ckeditor.dev/4.4.0/full-all/ckeditor/samples/plugins/placeholder/placeholder.html
2. Put caret in anchor "CKEditor"
3. Create some placeholder in caret position by clicking placeholder button located in toolbar
4. Hover over newly created placeholder
5. Click placeholder drag handler

Actual result: In WYSIWYG area there is loaded page with URL set in related anchor tag.

1. Open code snippet sample.
2. Focus editor.
3. Start navigating with up/down keys.

Reproduced in IE9 and IE11. I set version to 4.3, because since then we support widgets.

I've noticed it in IE8 and IE9. Didn't reproduce it with IE11 though.

Chromium and FF seems to be untouched by this issue.

1. open any sample with CKEditor and widgets (i.e. samples/plugins/codesnippet/codesnippet.html)
2. move your mouse to top border of the widget
3. click left mouse button

Expected result:
I think that nothing really should happen, i mean - focus should remain on widget.

Current result:
Caret goes into a drag handler.

additional info:

1. In IE9 caret sometimes went to the last widget on the page
2. I tried to reproduce it with captioned image2 / mathjax but none of these succeed, is it possible that it's codesnippet issue?

Browsers: All

1. Open magic line sample: ​http://presets.ckeditor.dev/4.4.0/full-all/ckeditor/samples/plugins/magicline/magicline.html
2. Set following content into editor

<p></p> <hr /> <hr />

3. Try to insert line between hr tags

Actual result: Magic line is not displayed in proper position

Please note: when add more hr tag, then magic lines are displayed between some of them.