Custom Query

The __NORICHEDITOR__ magic word could be created to disable FCKeditor on specific articles (those were too much Wikitext hacks are used for specific need).

I don't think it's really a bug, but i sure think it should be solved someday.

I was trying to improve the loadingtime of fckeditor, since in a application I use 14 plugins. Studying the logfiles of my server i discovered that fck_editorarea.css is loaded two times per instance (in that specific CMS I use five instances, which makes 10 requests for fck_editorarea.css). With an empty cache the server gives a '200' at the first two request, at the other eight request the server generates a '304'. Well, although a '304' doesn't generate traffic, it still does take some time.

So i think it would be better that fck_editorarea.css is only asked for once per instance, instead of twice. I was looking at the code, but could not figure out why it's called two times.

By the way: this only happens with IE 6 and IE 7.

Regards, Koen Willems

It would be nice to have an option to disable FCKEditor custom Context menu. This way, browsers spellcheckers (google toolbar or FF2) may be used, for example.

It may be done in fckeditor.htm, line 103:

// Initialize the editing area context menu.
FCK_ContextMenu_Init() ;

replaced by:

// Initialize the editing area context menu.
if ( FCKConfig.ContextMenu != null )
  FCK_ContextMenu_Init() ;

in fck_gecko.js, line 62:

// Reset the context menu.
if ( FCKConfig.ContextMenu != null )
{
  FCK.ContextMenu._InnerContextMenu.SetMouseClickWindow( FCK.EditorWindow ) ;
  FCK.ContextMenu._InnerContextMenu.AttachToElement( FCK.EditorDocument ) ;
}

and in fck_ie.js, line 129:

// Reset the context menu.
if ( FCKConfig.ContextMenu != null )
  FCK.ContextMenu._InnerContextMenu.AttachToElement( FCK.EditorDocument.body);

You just have to set FCKConfig.ContextMenu = null; in your custom config file to disable the custom menu in the editor area. Default context menu is still disabled on FCK toolbars. Most features are still available through toolbar buttons.

Proposed implementation tested on FF 2.0.0.3 and IE6.

It would be great to have a lite version for the Fckeditor, just for small needs like bold, italic, numbering, bullet, link. Cause right now, the control takes too much resources on loading.

IE is not working correctly when pasting a list of text.

1. On test site with IE, paste the following list:

* Escape from "Corporate-Cubical Hell"
* A Series of Law-Lectures
* The Weather
* An Overnight Jazz D.J.
* A 6-Pack on my Patio

2. Highlight text and hit bulleted list toolbar item.

3. Only a single bullet is created on the line above the Escape... line

Source code generated:

<ul>
    <li><br />
    * Escape from &quot;Corporate-Cubical Hell&quot;<br />
    * A Series of Law-Lectures<br />
    * The Weather<br />
    * An Overnight Jazz D.J.<br />
    * A 6-Pack on my Patio</li>
</ul>

Same behavior is pasting with Ctrl+v, Paste toolbar, Paste as Plain Text, or Paste from Word (with Ctrl+v)

FF generates a bulleted list using the steps above. Source code:

<p>&nbsp;</p>
<ul>
    <li>* Escape from &quot;Corporate-Cubical Hell&quot;</li>
    <li>* A Series of Law-Lectures</li>
    <li>* The Weather</li>
    <li>* An Overnight Jazz D.J.</li>
    <li>* A 6-Pack on my Patio</li>
</ul>

Implement the source view as an IFRAME with designMode=on (using FCKEditingArea), instead of using the <textarea>.

This change would give us many new possibilities to enhance the source view, like source code coloring, automatic selection positioning, and other features that could be implemented by using the DOM only.

Before doing this, we need to discuss all pros and cons of it, and the real possibility to implement it on browsers.

Yahoo Mail has a wysiwyg editor for composing emails, and its toolbar has only one icon for alignment which drops down for left/right/centre/justify. same with bullets. Also, the colour selector, font, font size, and smilies all use the same system.

Its doesn't overwhelm the user with buttons. simple like plucking fruit from a tree!

There's another sourceforge project which does the same little dropdown windowing system: ​http://sourceforge.net/projects/xsdheditor/ Not fckn half as good as fckeditor tho.

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=863562&group_id=75348&atid=543656

I just stumbled on this the other day: ​http://nanotree.sourceforge.net

Using this as folder tree would make for an excellent file and image explorer allowing users to categorize files and images using folders. Logically the file and image upload components would also have to use it to allow users to specifiy folders.

Initially webmasters would have to specify a folderstructure (or it auto-scans) and down the road users could themselves add/modify/delete folders.

Unfortunately I don't think my dhtml is up to par to do it myself.

This would make and already great editor excellent.

Please post your thoughts.

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=954899&group_id=75348&atid=543656

Following situation:

Prerequisites

• developer runtime (uncompiled scripts)
• fckstyles.xml

<Style name="Fat link" element="a">
    <Attribute name="class" value="aBold" />
</Style>

1. there is nothing more in the editor then 2 plain words without any formatting
2. select one word and create a hyperlink using toolbar "Insert/Edit link"
3. enter some target as usual, the hyperlink is created
4. now while the text is still selected choose the "Fat link" from the toolbar style pulldown menu

IE behavior

crashes with

Unknown runtime error
Line 28 in fckstyledef_ie.js > e.innerHTML =
oRange.htmlText ;
May be it is a IE bug...

FF behaviour

Firefox has 2 conditions

1. if the text is still selected it messes up the code

New document -> New <a class="aBold"><a
href="bala.ccc/">document</a></a><a
href="bala.ccc/" />

2. if the word has been re-selected it is ok (problem with selection cache?)

In both browsers if the style choise is done first, there are no problems.

This problem is not as big in new documents as in old stuff where hyperlinks exist and have to be "re-styled" or modified.

Best regards

SelfMan

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=1249695&group_id=75348&atid=543653

in "FileWorkerBase.cs" -> "UserFilesPath" property there is a mistake in order to determine which of "sUserFilesPath" variable will be set. Watch this:

[original code]

...
...
// Otherwise use the default value.
if ( sUserFilesPath == null || sUserFilesPath.Length
== 0 )
sUserFilesPath = DEFAULT_USER_FILES_PATH ;

// Try to get from the URL.
if ( sUserFilesPath == null || sUserFilesPath.Length
== 0 )
{
sUserFilesPath = Request.QueryString
["ServerPath"] ;
}
...

code

spotted? "Try to get from the URL." never works.

[new code]

...
...
// Otherwise use the default value.
// Try to get from the URL.
if ( sUserFilesPath == null || sUserFilesPath.Length
== 0 )
{
sUserFilesPath = Request.QueryString
["ServerPath"] ;
}
if ( sUserFilesPath == null || sUserFilesPath.Length
== 0 )
sUserFilesPath = DEFAULT_USER_FILES_PATH ;
...

code

Greetings, Gürhan Başbuğ.

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=1368066&group_id=75348&atid=543653

It would be cool to develop custom palettes in FCKConfig.FontColors and then be able to assign those palettes to specific ToolBarSets. I imagine that you could apply this paradigm to other similar items (FontNames, FontSizes, etc.) Perhaps by adding another layer of abstraction via an array.

"smiley" is regarded as unprofessional in business use.

The more generic term "icon" would be less problematic and mean the same.

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=1274859&group_id=75348&atid=543656

Check out ​http://www.tiddlywiki.com/. This personal wiki program uses JavaScript to save files locally. It only uses html and javascript to do this. It works for all the major browsers (firefox, ie,...). I would like to see your program use the same type of technique to save it's data.

Moved from SF:
​http://sourceforge.net/tracker/index.php?func=detail&aid=1401907&group_id=75348&atid=543656

Pressing the Enter Key after a <li>&nbsp;</li> will create a new <li></li> instead of removing the <li>&nbsp;</li> that seemed empty for the user.

In the line of thought my request 'create toolbar(set) at runtime' (Ticket #937), it would be great to be able to create buttons at runtime. I mean completely new buttons that were not available as plugins.

The uploader in PHP contains multiple instances of

chmod( $sFilePath, 0777 ) ;

for uploaded files or directory creations.

Can I suggest this be changed to 0774. I can't think of any reason why the very insecure o+w should be enabled. 0770 would be preferable, but I guess there could be webservers that have the webserver process not in the main user/group.

Using fckeditor with BodyClass='cpage'; define in main html: .cpage h1 { ...any-style-here... } when editor loads, all the format preview tabs are blank (white/white text) except where specifically defined in css with color:black

add online editing, allowing multiple users to redline and display all edits to the ower

I know that people should read docs etc. but let's see how some of users test software before actually deciding about using it. They download file, unpack, then see _samples dir and simply run examples. Since there is no README in _samples dir, it's not obvious that to run examples divieded into separate dirs for each language, user must do some additional config changes (that's how examples usually work).

So, because we can't split FCKeditor into separate distibutions for each server language (which would give us real statistics of using each distribution btw.) I think it's important to give at least the most appropiate and descriptive error responses to the user to that specific error (and generally everywhere).

Very good example is problem with wrong _FileBrowserLanguage and _QuickUploadLanguage. In error message we are displaying contents of a whole file and "The server didn't send back a proper XML response." (see #227) - it says nothing to the user and it's not so easy to close that message on a small screen.

I suppose that some people wouldn't consider it as a configuration error but as a shortcoming and would be wondering whether they encounter more strange problems in the future.

My suggestion is that we could add in each connector.ext file some special tag in a comment section e.g.

do not remove this line: #FCK_WRONG_LANGUAGE_ASP#

And then if XML response is wrong, we could check before displaying error message whether such special tag exists and display more friendly information, for example:

Your server didn't execute ASP file. 
Please check whether ASP is available or change your default server language in fckconfig.js (...) 
You can find more information here:...

There are security settings on browsers to block scripts from disabling the default context menu. In those cases, users will have no way to display our custom menu, as the only option we have for now is the right-click (as expected, btw).

We should provide alternative ways to display the menu. Some proposals are:

• [Left Click] + [Keyboard Accelerator] : we could use CTRL on Windows and META on Mac. The problem is that IE uses CTRL+Click to select the paragraph. I'm not sure we have people using that our there, but we would be removing that feature from the browser. ALT instead is used to activate the application menu on Windows, so we may have conflict.

• CTRL + [Space] : we can also make it configurable in the fckconfig.js file. It means that a "ContextMenu" command should also be created.

Scenario:

1. Create a table
2. Click the mouse outside of the table (either just before or after)
3. The cursor is placed outside of the table at the height of the last row
4. When starting to write the cursor is moved to the correct position (either into the paragraph above or the paragraph below).

Tested in nightly build with IE7

The wrongly placed cursor is a bit confusing for the user.

Dear Fred & developers,

There'd better be drop-down/combo box for dynamically switching new skins (Default/Office/Silver). We could add cookie implementation for remembering user's favorite skin. Currently what we see in JavaScript Sample 05 is a separate part of editor toolbar. What I'd like to suggest is that drop-down box combined in editor toolbar.

Thank you so. aungkhant

Dear Fred and developers,

In FCKeditor, the toolbar plays a vital role. What about Menu? Yeah, menu like 'File', 'Edit', 'Search', 'View', 'Format' as we see in windows applications.

The menu will be at top of the toolbar or a temporary replacement of the toolbar. I imagine it'll be really nice to see both Menu and Toolbar together like in Word application.

Thank you so much for your time, consideration and hardwork.

Dear Fred and developers,

As you already know, CSS can do incredible styling that plain HTML cannot.

The newly CSS-Style button is disabled by default. When the user types some text, it becomes enabled.

Click it. The "CSS Properties" dialog will appear. The dialog will have server tabs of CSS categories :-

1. Text
2. Background
3. Box
4. Block
5. Border
6. List
7. Positioning

Each category will have their respective CSS elements. e.g. For Text, there will be -

1. font-family
2. font-size
3. font-weight
4. font-variant
5. color
6. background-color ....etc

The eraser icon named 'Remove Format' must be able currently applied CSS style.

Thank you so much for your time, consideration and hardwork.

Steps to Reproduce

1. Open sample01.html and enable "Show Blocks";
2. Hit Enter at the end of the paragraph;
3. Hit Shift+Enter.

The cursor will make a "quick strange blink", but there will be no visual results.

Switching to source view, the second paragraph will display the correct HTML:

<p><br />
&nbsp;</p>

So it seams that the Shift+Enter itselft is working, but the selection is not being made visible after the <br>.

This ticket is not here to define our definitive approach for it, but to open a discussion around this argument.

With the recent rework made for the paragraph alignment features, we have also introduced an "automatic default" for the alignment buttons. It means that if the content is LTR, the left align button will be enabled by default. In RTL instead, the right align button will be highlighted instead.

It sounded nice in the beginning, but now I'm a little bit worried about it. Let me bring some thoughts so we can discuss about it.

Let's suppose we have the following HTML:

<p>This is a test.</p>

Looking at it as is, what is the alignment of this paragraph? "None", is the only correct answer. But FCKeditor says it is "Left Aligned" if LTR or "Right Aligned" if "RTL".

The document direction setting in FCKeditor are there just to help reflecting the final result we'll have when loading the contents at the target web site. But the fact is that the direction information is not in any way saved along with the HTML produced with the editor.

Now... what if I want to "force" the paragraph to be left aligned, even if the document is LTR. Here is a simple sample:

<td align="center">
  <p>This is a test.</p>
</td>

Here the editor is saying that my <p> is left aligned, which is not true. It is actually has no alignment specified, so it simply inherits it. But I can't make it left aligned, because the editor is saying that it is already to the left. There is a hack for it: align to right, and then to left.

Those are just some issues that came to my mind. I know MS Word has this default behavior, but is also true that MS Word is not made to produce pieces of content to be injected, as FCKeditor does. Who knows how creatively will the content be used by our developers base.

Maybe I'm wrong, so please correct me in this case.

switch to source and paste

<div id="black_box"></div> 

switch twice and you'll see that the content has been lost.

(any browser)

Reported in ​http://www.fckeditor.net/forums/viewtopic.php?f=6&t=7251

When the cursor is placed next to (as IE puts it) a "control" element, such as an image or a table, it should select it on delete / backspace.

This improves the intuitiveness of deleting and makes sure people dont delete things accidently.

Here is a function partially implemented which does this

FCKListsLib.ControlElements = {'img':1, 'table':1, 'input':1};

// Helper function upon deleting a node that may be a control (must select it first)
// A control is something like an image or table which can be selected with the cursor
FCKEnterKey.prototype.HandleControlNodeSelect = function(pRange, bIsForward)
{
	// Check to see if we are about to delete an image or table
	// TODO: this is IE specific at the moment!
	if(FCKBrowserInfo.IsIE)
	{
		// TODO: this only works when a element is in a block
		var pStartBlock = pRange.StartBlock;

		if(pStartBlock)
		{
			var iDirection = bIsForward ? 0 : -1;
			var iOffset = pRange._Range.startOffset;
			var pPrevious = pStartBlock.childNodes[iOffset + iDirection];

			if(pPrevious && FCKListsLib.ControlElements[nodeGetTagName(pPrevious)] != null)
			{
				FCKSelection.SelectNode(pPrevious);
				return true;
			}
		}
	}
	
	return false;
}

Split from ​http://dev.fckeditor.net/ticket/1325#comment:10

I don't think that it's used, so it's one item less in the configurations.

Using ie. div in EnterMode produces a clean div tag that embraces its content upon pressing enter or when loading the editor. Ie. <div>New paragraph</div>

Could it be possible that the div-tag created by EnterMode to be affected by DefaultStyleLabel and/or CoreStyles?

For instance, if having

FCKConfig.EnterMode = 'div' ;
...
FCKConfig.DefaultStyleLabel = 'myClass1' ;
...
FCKConfig.CoreStyles = 
{
...
'div' : { Element : 'div', Attributes : { 'id' : 'myDiv1' } },
...

the EnterMode currently produces:

<div>Text</div>.

When it really should produce

<div class="myClass1" id="myDiv1">Text</div>

It would be nice to have the ability to define the attributes the default block element should have. We are using 'p' as entermode and only classes to define the different styles in the text.

The customer have the option to define the default paragraph style. It would have been nice if every methods instead of creating a empty paragraph tag created a paragraph with default attributes (e.g. default classname).

Browsers: Firefox 2.0.0.11 (win+lin), IE6+7

1. Go to ​http://www.fckeditor.net/demo/
2. Open source view
3. Press ctrl+z

What happens: All the content is removed.

What I expect: Only changes that I've done to be removed, not all the source.

You'll have to use ctrl+y (or ctrl+shift+z) to get the content back because the undo/redo buttons in the tool bar are deactivated. Personally I think that's ok, but several users claim to know about the ctrl+z shortcut and not the redo shortcut.

If a page has a category entry, ex: category:animals, if you edit the page and type over the old category name with a new category name, such as reptiles, the category name does not change when you save the page.

This extension should have a save and cancel button in the toolbar!

Would it be possible to configure the activation of xhtml validation in the configuration file. This would resolve the problem raised in ticket 252 : ​http://dev.fckeditor.net/ticket/252. I have the same issue (Internet Explorer 6 & 7).

The syntax generated in source view for IE is missing a semicolon when selecting a style or size format.

1.) Hightlight some text 2.) Select a custom style or font size (Highlight Green or Small for example) 3.) View source and take not of the syntax. <span style="font-size: small"> It should be <span style="font-size: small;">

This happens in IE 6 and IE 7 that I have noticed. It outputs correctly in Firefox.

User must be able to click on the link in editor area without holding CTRL button to open the linked page. Normally, in the current scenario, if user creates a link in FCKEditor, then to open that link, the user has to click on that link by holding CTRL key. What I'm asking is that user must be able to open that link page directly by clicking on it without holding CTRL key.

Thank you.

A sidebar that can contain toolbar buttons and/or plugins/widgets. E.g. for often used toolbar buttons. Giving them a bigger button in a sidebar will ease the use of them. Or for often used plugins or plugins that want to display a little bit more, e.g. a clock that onclick displays a date picker.

Example: Opera.

A handle at the bottom border of the editorArea or in the right-bottom corner to ondrag resize the editorArea.

Being able to benefit from javascript frameworks within FCKeditor would be nice. But with everyone having his favorite one it should be possible to use multiple frameworks without having them collide with each other, or with the FCKeditor itself.

Perhaps this can be done by running each one in its iFrame and providing Objects through which each framework its method can be used? Not sure if there something available like this already, but maybe ​Mootools' slickspeed of which ​the source is available, can be used as a guide?

An ajax framework within FCKeditor that is easy to use for plugin developers.

FCKeditor as en engine so that any custom GUI can use the functionality of it...

See ​forum thread.

When the editor becomes ​more plugin based it might be a good idea to make the edit modular, so that people can download exactly what they want. A download page where functionality and/or plugins can be selected could be provided. (See for example the ​mootools download page with selectable compression type).

See ​forum thread.

(voting for core functionality could be an alternative)

A table wizard as a GUI to create (complex) table structures.

See ​rad editor's table wizard

I thing it's a good idea to use the ​Tango icon set by default. As it states on the website, the Tango Desktop Project exists to help create a consistent graphical user interface experience for free and Open Source software. I believe this is a great concept and one that FCKEditor can really benefit from.

On the Tango site there is only a basic set available, but I believe other icons are available as well and can be found by searching with google ('tango' + 'openoffice', or the name of any other open source software for example).

See the ​graphical design V3 guidelines

See the ​forum thread.

I have an implementation that allows for webpage editing using WYSIWYG. When the view is switched to "Source" the cursor is positioned to the beginning of the page forcing the user to manually scroll to the area of interest.

I would like to have the cursor position maintained eliminating the nuance of manually scrolling. For this I need a pointer to the element that holds the editor text.

See the following code to obtain the cursor position on IE:

<HTML>

<HEAD>
<SCRIPT LANGUAGE="JScript">
function saveCaret(elem)
{
  if ( elem.isTextEdit ) 
    elem.caretPos = document.selection.createRange();
}
function getCaretPos(elem)
{
  if ( elem.isTextEdit && elem.caretPos )
  {
    var bookmark = "~";
    var orig = elem.value;
    var caretPos = elem.caretPos;
    caretPos.text = bookmark;
    var i = elem.value.search( bookmark );
    window.status = "Caret is at character " + i;
    elem.value = orig;
  }
}
</SCRIPT>
</HEAD>

<BODY>
<INPUT NAME="txtInput" ONSELECT="saveCaret(this)" 
   ONCLICK="saveCaret(this)" ONKEYUP="saveCaret(this)" VALUE="Where are you?">
<INPUT TYPE="button" VALUE="caret pos" ONCLICK="getCaretPos(txtInput)">

</BODY>
</HTML>

Can anyone tell me how to get a handle of the element that has the text in the Editor?

If you have one or more styles set in fckstyles.xml which use the p element, in the editor every <p> shows as having the styles selected, even if they are not. You are therefore unable to set the p style on a paragraph.

Example styles in fckstyles.xml:

<Style name="Intro Text" element="p">

<Attribute name="class" value="IntroText" />

</Style>

<Style name="Sub-Head" element="p">

<Attribute name="class" value="SubHead" />

</Style>

Inline popups are very great in the last beta; unfortunately inline popup are not applied for file browser. Is it possible to switch from open.window to FCKDialog.OpenDialog ?

It seems that it's too hard for the people to read the documentation or understand how their server works, so there are one post after another in the forums about people that try to install the editor in Asp.net and forget about all the js files or place them in the wrong folder: ​http://www.fckeditor.net/forums/viewtopic.php?f=6&t=8312 ​http://www.fckeditor.net/forums/viewtopic.php?f=5&t=8698

There are more threads like that, one of them seemed to say that the BasePath parameter wasn't obvious.

Some possible ideas:

• From the asp.net code, check that the FCKeditor folder pointed by base path does exist.
• Offer to download the code (if it isn't found, on first run... I don't know).
• Provide a single package for the lazy people that contains both packages.
• Make the BasePath default to ~/FCKeditor

I don't know about you, but in my mind splitting a cell that hasn't been previously merged just looks strange.

By restricting those operations only to cells that have colspan or rowspan >1 we will avoid ugly tables and the code will be very easy to manage, we just need to decrement the colspan or rowspan value, there's no other things to worry about.

Steps to reproduce

1. Type a line into the 2.6 beta

2. Bold that line

3. Hit return from the end of that line

4. View source

Expected result: <p><b>&nbsp;</b></p>

Observed result: <p><b></b></p>

Consequences: The line you intended to add after the bolded line will not appear when the page is saved and viewed.

Note: This occurs with other style tags as well.

Standard ticket submission mantra: I searched but did not find, if I erred, the fault is mine.

These are more layout requests rather than bugs... Thought I'd submit it just incase you think they are good ideas...

Do you think it is a waste having a whole tab dedicated to the target of the URL? On the Link tab for the image dialog, you have the target on the same page. This will save a click every time you insert a link you want to open in a new window...

Same with the Advisory Title. Most of those Advanced tab features people wouldn't use but they are worth keeping. However, the Advisory Title is something that is used quite commonly for usability and especially SEO... It might be good to put that in the same tab as the URL as well. I for one use it all the time.

Lastly, can we add a Hyper link Relationship text box (rel="")? This is used in a lot of javascripts these days such as ligthbox (<a href="photo.jpg" rel="lightbox">) and can also be used for SEO. This can go under the Advanced tab as it wouldn't be used that often but is a handy feature.

If you are greeted with the following error when trying to upload files using the file manager:

Using themed css files requires a header control on the page. (e.g. <head runat="server" />).

You don't need aspnet themes within the fckeditor folder, so simply add a web.config file in the folder, with the following content:

<configuration>

<system.web>

<pages styleSheetTheme=""> </pages>

</system.web>

</configuration>

Port the file browser from V2.

The default editor installation already proposes some default styles to be used in the editing area. Those can be found at the contents.css file, which should be the file to get customized on editor installations, to reflect the final web site styles, giving a more WYSIWYG feeling.

Right now, we have included just font face, size and colors. But we could include further styles in that file by default, so creating text with the editor will look nicer inside of it, and would be more usable.

One of the example of it is the <blockquote> tag, which currently looks just like an indented paragraph. It would be nicer to have it look differently, helping understanding that that block is a quotation (#3179), and that it's different than a simple indentation.

There are other cases, like less used tags, like <kbd> and <code>, which could have better styling.

We do that with FCKeditor at sample14.html, with good results. Maybe it's the case to do that by default. After all, that CSS file must always be configured, so if one wants, s/he can just remove the unneeded rules.

Reproducing Procedures

1. Open the replace by class example page;
2. Make the following content with the selection;

   <table border="1" cellpadding="1" cellspacing="1" style="width: 200px; height: 46px;">
   	<tbody>
   		<tr>
   			<td>
   				^text
   			</td>
   			<td>
   				text^
   			</td>
   
   		</tr>
   	</tbody>
   </table>
   

3. Click 'Indent' button;
   • Actual Result:

     <table border="1" cellpadding="1" cellspacing="1" style="width: 200px;">
     	<tbody>
     		<tr>
     			<td>
     				<p>
     					text</p>
     			</td>
     			<td>
     				<p>
     					text</p>
     			</td>
     		</tr>
     	</tbody>
     </table>
     
     

• Expected Result:

  <table border="1" cellpadding="1" cellspacing="1" style="width: 200px; height: 46px;">
  	<tbody>
  		<tr>
  			<td>
  				<p style="margin-left: 40px;">
  					text</p>
  			</td>
  			<td>
  				<p style="margin-left: 40px;">
  					text</p>
  			</td>
  		</tr>
  	</tbody>
  </table>
  
  

Currently the templates plugin only allows to load the data from a js file without an option to use a XML like v2.

This has some drawbacks:

1. It's not possible to reuse existing templates. Asking the people to rewrite their templates can be a big issue as it might involve other tools that they had setup to provide custom templates for each client.
2. A json file is more complex to write correctly than a XML. To verify that a XML file is valid they just have to open it in the browser, and it will point to any error. Validating the js file will be more complex and I guess that most of the people will follow the trial and error approach, leading to frustration as now they have to learn the javascript rules in order to make it work.
3. Managing XML files seems easier to me. I think that any framework has tools to read/write xml files, merging them, etc...

So I suggest to improve the templates plugin allowing to load templates from XML files with the syntax from V2.

This's a ticket derived from #705, which provide a UC of the following style definition which we don't support now:

{ element : 'strong', overrides : [ { element: 'b'}, { element: 'font', styles : { 'font-weight' : 'bold' } } ] }

And a more generic UC ( and actually a common case )with 'style' only definition could be inferred as:

{ element : 'strong', overrides : [ { element: 'b'}, { element: '*', styles : { 'font-weight' : 'bold' } } ] }

Which declare that we need to consider bold for the following two cases:

1. It's a <strong> tag;
2. It's a <b> tag;
3. It's any tag with "font-weight" : "bold" style;

Reproducing Procedures

1. Open the replace by class example page;
2. Make the following selection;

   <p>
   	<strong>bo[ld</strong> and lin]k</p>
   

3. Open link dialog to add a link;
4. Make a new collapsed selection as following and click 'Unlink';

   <p>
   	<strong>bo<a href="http://example.com">ld</a></strong><a href="http://example.com"> and ^lin</a>k</p>
   

• Actual Result:

  <p>
  	<strong>bo<a href="http://example.com">ld</a></strong> and link</p>
  

• Expected Result:

  <p>
  	<strong>bold</strong> and link</p>
  

Now if we have the following content:

line1
<br />
<div>
	line2<br />
	line3<br />
</div>
line4

It's not possible to 'free' line2 and line3 outside of the wrapping div in wysiwyg way.
One proposal is introduce a new Format/Style named 'Body Text'

which remove the block style from selection.

The Problem

The current filter system doesn't respect filter rule priority:
When rule A is specified with priority higher than rule B when adding to filter system, what expected is all rules items of A will be applied before rules from B.
Unfortunatelly, the current system doesn't allow this happen: the current behavior is different, in which it only guarantee the priority of the same rule item, so one rule imposed on the parent node could never come later than it's child's rule.
If now we have

• a rule entry in A, said, e.g. remove all empty <p>;
• a rule entry in B, said, e.g. remove empty <a>;

Then there's no way to guarantee both rule's completeness.

The Solution

The reason is due to html filter is tightly binding to writer now, which does not allow it to run rules in multiple times, the rules is applied along with the writing process.
If we isolate the filter system out, it becomes more flexible:

1. Allow "parsing once, filtering at will", with each filter rule changing the lightweight dom tree structure on the fly;
2. Allow filter to alter node type easily;
3. Write the whole document after all filters have applied.

Currently once bookmark is created, broken text node will be forever stay in pieces, which result in a fracted document, have negative impacts on dom walking performance.
For this reason, possible treatment should be done in CKEDITOR.dom.range::moveToBookmark to merge sibling text nodes.

URL: ​http://ckeditor.com/ckeditor/3.0rc/_samples/skins.html
OS: Windows XP SP 3
Browser: K-Meleon 1.5.3 ​http://kmeleon.sourceforge.net/

See attachment.

Hi,

First of all I wish to thank all those who have contributed to this tool. It works amazingly well. I'm using FCKEditor 2.6.4.1 for my project work. I'm using the following software for my project: 1) JavaScript, HTML, 2) Operating System: Linux, 3) Browser: Firefox. I have integrated this editor to my existing project. It is working great. However, I was asked to add new features to it. I need paragraph, sub-paragraph like structure. To be more clear following is the kind of structure I need to create:

*

1. 2.

a. b.

(i) (ii)

aa. ab. ac.

(iii)

aa. ab.

(iv)

c.

(i)

aa.

(ii)

d.

a.

(i)

aa. ab.

(ii)

b.

(i)

4.

*

The structure has to follow this pattern only. For paragraph it starts with 1. For sub-paragraph it starts with a. For sub sub paragraph it starts with i (small roman numeral). For sub sub sub paragraph it starts with aa. I'm actually able to create this structure. However, whenever I need sub paragraphs I have to right click and set the ordered list element's "type" attribute and "start" attribute to its appropriate value.

I need a feature using which we get the above pattern when the user presses "indent" button or "tab" key (whichever is convenient).

Again, I'm using 2 such editors on the same page. My requirement is that first editor should start with paragraph number 1. We must then keep track of the number of paragraphs (in the above example it is 4) entered by the user in the first editor. Then the paragraph number should continue from the first editor. So in my example the second editor should start with number 5 when it receives the focus or when the user starts entering the value.

It would really help me if you could tell me how I can modify the editor to get the desired result. I tried to learn the code. But since I'm a beginner I feel that the code for this editor is very vast and I'm unable to trace the necessary code.

I would appreciate if any further help is provided on how to change the code and what are the files to access for this purpose.

Thank you all once again for this editor and for your help.

It would be useful to have the preview mode (just like Source).

1. empty the document
2. insert a table
3. try to place the cursor before the table (to enter some content)

you can't

This problem applies even to CKeditor 3, where you are adding a dummy <p> before the table apparently to workaround the problem. If this <p> is removed, the user gets in the same problem.

See #4122 for a concrete example of it, though we've already provided a temporary at that ticket, a general solution should be found.

#3658 has been using the same hack from v2 to eliminate horizontal document scroll with xhtml doctype by forcing the vertical scrollbar, which is obtrusive for a document need no scrolling at all, a better solution should be found.

SUP and SUB script should be turn off after newline.

TC

1. Use following content with selection:

   <p>
   	<sup>sup^</sup></p>
   

2. Press enter
3. Type any chars - there are still superscript'ed.

CSS selector support for Test Environment would bring us such benefits for node-related TCs:

• shorter, leaner, more readable TCs
• less time needed to implement a TC
• more general node-access code formulas (possibly reusable)

"Node-related" TCs means any operation which operates on specific nodes (in both forms - as plain text and DOM representations) which are placed in larger set of other nodes. This can include things like:

• click specific button
• check if results inside editor is the desired one
• add new markup in specific place

Possible solutions for this could be provided by:

• ​sizzle
• ​DOM Assistant
• ​prototype (which ships with CKTester from #4218)
• ​YUI selector (which could share it's core with YUI tests)
• ​dojo.query

One popular selector benchmark is ​slickspeed (although speed isn't out concern).

I think that most important factors when making a decision are (for us):

• size of additional code
• stability across all browser (including IE6) in dangerous TC environment
• support level

Today it is possible to send css files as an array in the definition given to CKEDITOR.ui.panel constructor.

In my case it would be very nice if I could send the style as a string since the css is already loaded on the page and it is calculated by php so the file is not cached by the browser.

My suggestion is to add a config option to the panel class called "style" or something that is applied to the created iframe inside a style tag.

Currently it's been difficult to find the relationship of UI instances ( e.g. buttons, combos etc. ), and it's related toolbar instances. Adding more cross-reference among them could benefit the plugin authoring which will manipulate the toolbar.

CKEditor behaves in a different way than FCKeditor when height is set to percentage value.

Suppose we have a floating resizable window, where the editor is running. When height of the editor is set to 95% and window is resized, CKEditor does not change its height. FCKeditor adjusts its height to match the new size of the window without any problems.

CKEditor can be easily changed to work in a similar way as FCKeditor, by changing the CSS style to:

.cke_skin_kama .cke_editor{
  display:inline-table;
  width:100%;
  height:95%; /* < --- added */
}

in skins/kama/editor.css

.. but this way the "resize" plugin stops working.

Custom plugins and extensions are able to extend the list of loaded styles in the styles plugin. Unfortunately the styles plugin does not reflect dynamic changes to its list of styles and requires a reload of the whole editor to reflect the new options.

Ideally the editor should provide a way for any plugin to be reloaded without requiring the whole editor to be reloaded. Alternatively, a simpler approach to this problem would be to allow just the styles plugin to reload its data dynamically, through some additional API.

Hi,

I think menu buttons width is bad.

When changing language in 'fr' you can see the bug in attached image.

Thx

Jean-mat

Hi

Right click menus width is fixed, and in other languages it's a problem. See the attached file.

Best regards

Jean-mat

Mozilla Lab Bespin Embedded just offered a ​preview released, which is by nature a excellent 'source' editing block alternative for us.

A 'bespin' plugin should be created to start the integration effort with Bespin though the current release is still missing many important features for us.
Bespin DOEST NOT support all versions of IE and Opera at this moment.

When Remove Format is applied to block level headings, they are not reduced to the default text.

When custom styles such as "Blue Title" are applied to some text and then the Remove Format feature is used, the color from the style is removed, but the heading is not. This seems inconsistent.

This combination encode text in editor.

1. In samples config.js set config.htmlEncodeOutput = true
2. Go to jQuery adapter example
3. Click Submit
4. Go back
5. Editor display encoded html

I'm not sure if this is intentional/required - however it makes sense to me to have the following files in the '_source' directory, since they are source files.

ckeditor_basic_source.js ckeditor_source.js ckeditor.pack (would obviously require changes within for the paths)

It just makes it easier to manage all source files, as I wont be including these source files in my projects package - deleting just 1 directory is easier.

Looking at the code in plugins/undo.js I've found two statements like

	return this.restoreImage( image ), true;

That ", true" is useless as far as I understand and the lint check doesn't give a warning because it's disabled in its config:

-comma_separated_stmts        # multiple statements separated by commas (use semicolons?)

Enabling that shows that there are other places where such problem exists. Some of them I guess that can be fixed by changing the comma to semicolon, but I don't know what's the logic for other situations like this one in tabletools:

	if( trimCell( cell ), cell.getChildren().count() )

maybe it's a && there?
But as that can change the behavior each change should be done knowing what's the logic that should apply.

Create an editor instance that spans the with of the screen.
Create a table with 100% width.
Move the mouse to the right hand side of the editor and right click on the table.
Observe the context menu correctly displays as fixed in #4594.
Open a sub-menu, e.g. Cell Properties.
Observe that the sub-menu covers the main context menu.

Hello,

During my tinkering with ckeditor, I think it would be very convenient to have the methods "is()" and "getName()" moved from element.js to node.js.

This makes sense because nodeName is a part of the browsers TextNode object as well. It helps because if you are doing a test to see if an element is say a <br /> the code changes from:

if(element && element.type == CKEDITOR.NODE_ELEMENT && element.is('br'))
or:
if(element && element.is && element.is('br'))

to just:
if(element && element.is('br'))

I have seen these sort of checks a lot of places in the ckeditor source. Let me know your thoughts.

This is due to the patch in #4973, but it was a long talk that would distract from the real patch.

The patch in #4972 (v2) removes CKEDITOR.loadStylesSet and CKEDITOR.addStylesSet, so when people tries to upgrade they will get an error if they have used them.

I think that most of the people didn't know anything about CKEDITOR.loadStylesSet unless they had to fight with the lack of CKEDITOR.stylesSet ​as I did, but I think that there's lots of people using their own styles with the addStylesSet call. If they don't change the code then it will fail and they have to find out the reason.

So my question is: how do we approach backwards compatibility?

We can leave the code as is and they have to read the release notes to notice the reason of the problem. This can be more complex if they are upgrading for example from 3.0 to 3.3 with hundreds of bugs in the mean time.

We can add bold and red statements in the what's new to make it clearer, but they still have to read the docs.

We can provide also a "compatibility" plugin that it's used just to provide compatibility with older apis. This plugin could define just the old functions as mapping to the new APIs and launch a silent warning in the console (if it exists)

Something along these lines:

CKEDITOR.addStylesSet = function( name, styles ) 
{ 
    if (window.console)
        window.console("The CKEDITOR.addStylesSet function has been deprecated. Please use CKEDITOR.stylesSet.add. Read... for further info");
    CKEDITOR.stylesSet.add(name, styles)
};

This way the upgrade is easier, there's a very little overhead and people can remove the plugin when they know that everything is working OK.

In both basicWriter and htmlWriter the openTag method doesn't use the attributes parameter:

		openTag : function( tagName, attributes )
		{
			this._.output.push( '<', tagName );
		},

and

		openTag : function( tagName, attributes )
		{
			var rules = this._.rules[ tagName ];

			if ( this._.indent )
				this.indentation();
			// Do not break if indenting.
			else if ( rules && rules.breakBeforeOpen )
			{
				this.lineBreak();
				this.indentation();
			}

			this._.output.push( '<', tagName );
		},

Am I missing something here?

In the calls to the method the attributes are sent, but that could be removed as well as the examples about how to use it.

To help with the usability of the context menu for JAWS users, the first item in the context menu should be selected (and read by JAWS). At the moment there are some issues understanding how to use the context menu by a JAWS user.

In the Find/Replace dialog, the button should be labeled "Close" and not "Cancel", because it doesn't really cancels the replacement operation.

In link\plugin.js there are two options that aren't documented

CKEDITOR.tools.extend( CKEDITOR.config,
{
	linkShowAdvancedTab : true,
	linkShowTargetTab : true
} );

As the new architecture of the dialogs allow to easily customize the tabs of each dialog, I wonder if we should add the documentation for these or just remove them (and the little code in the dialog.js file)

Personally, I vote for removing them. Note also that the image plugin lacks the corresponding options.

To reproduce the defect

1. Open Ajax sample

2. Click on Numbers icon to start a Numbered list.

3. Type some text after the first number in the list.

4. Select the text in list item and apply Pre-Formatted formatting.

5. See that Pre-Formatted formatting is applied to the list item.

6. Now press Enter

Expected Result

See that next number comes up in the ordered list and when we press the Enter for the second time,the cursor should come out of the Ordered list and a new empty Paragraph should open

Actual Result

The Next Number in the ordered list do not comes up and when we Press Enter any number of times we can't come out of the list.

Same behaviour is also happening in the case of Bullets.

Reproducing Procedures

1. Load any of the sample page and fill the editor with the following contents:

    <ol>
   	<li>
   		item1</li>
   	<li>
   		&nbsp;
   		<ol>
   			<li>
   				item2</li>
   		</ol>
   	</li>
   </ol>
   

2. Place the cursor at the beginning of the second list item (empty content) and press 'Enter' key;

• Expected Result:

   <ol>
  	<li>
  		item1</li>
  </ol>
  <p>
  	&nbsp;</p>
  <ol>
  	<li>
  		item2</li>
  </ol>
  

• Actual Result:

    <ol>
  	<li>
  		item1</li>
  	<li>
  		&nbsp;</li>
  	<li>
  		&nbsp;
  		<ol>
  			<li>
  				item2</li>
  		</ol>
  	</li>
  </ol>
  

Hi there,

When you are in a paragraph, indent adds margin. When you are in a list, indent adds nested list. When you are in a blockquote, indent adds...

1. a blockquote
2. indent a paragraph
3. indent the blockquote
4. do nothing

There is no best choice isn't it?

Browsers parse the contents of the style attribute and do strange things with its contents, so we should protect like we do for href and src so it's modified only when the user request it, and not by the browser.

Demonstration

• Open the demo and open the Image dialog.
• Go to 'Upload' tab and upload a file whose name contains characters that should be URL-encoded.
• After the upload is finished, sometimes a JS is thrown and the dialog does not switch to the 'Image Info' tab. in that case, switch manually to this tab.
• Take a look at the URL field. The characters that should be encoded (along with the space of the 'Public Folder' directory in the path) are not URL-encoded.

(For comparison, click on the 'Browse Server' button and select the file you have uploaded and notice that those characters are encoded)

People usually don't know anything about html tags, they just want things to work, so as long as it's possible we should try to think like a new user and how the expect the editor to behave.

In order to end a PRE-formatted paragraph I think that it's easy to expect that just selecting "Normal" in the format dropdown should start a new paragraph and get out of the pre if the caret is at the start of a new line. This would work in a similar way to the Bold button: you press to start marking things as bold and when you press it again you end the effect.

Provide a plugin to detect desktop file dropping into editor, for those support browsers.
The detected file info could be used to upload the file immediately or perform any possible customization.

Edit: Yes that is most correct. Editor should have at least a hook for implementing drag&drop into editor. What I mean is that if editor is connected with file uploader (like CKFinder) it should prepare image information that uploader can use to upload file. After file is uploaded, uploader should send new image path and editor should apply it.

Version 3.3 added support for preserving heading structure when creating a list item. Would be nice to also preserve structure when creating list items from paragraphs.

It is a known issue that IE does not properly allow for selecting content in multiple cells in a table to delete their content. The purpose of this ticket is to discuss whether CKEditor can do anything to facilitate the use case where a user selects text from multiple cells (using mouse or keyboard) and hits the delete or backspace key. The desired behaviour is that the selected content is removed from the cells. Firefox and Webkit handle this case natively.

Is it possible for CKEditor to detect the delete keystroke and perform the desired removal for IE? I realize this is working around the lack of functionality provided by IE, but it has come up as a customer request.

1. Create new paragraph.
   
2. Create new table.
   
3. Create new paragraph.
   
4. Select all content and click "Center Justify" button
   

In result alignment will be changed for paragraphs and table's cells. Expected: alignment should be applied for the table element.

Enter at the end of DT element should create new DD element and vice versa.

Unlink should remove only selected text for non-collapsed range.