I just want to do simple things like I used to do previously keeping a local repository of CKEditor and the patches that aren't being added to the main code. Using TortoiseSVN that was a breeze, no need to remember anything, just go to the explorer right click and under the Tortoise everything was available right there. No matter if I didn't touch SVN for months and I forgot everything about CKEditor and SVN, it's just a matter of reading the name of the options and then click Update after finding it. If the changes were too complex then its merge editor showed the problems and I could easily adjust the conflicting code.
Now CKEditor works with GitHub and so many other people also live and die by GitHub that I've been trying to use it, but it's so different that I can't spend the time to forget everything that I know and start over as some of the books on GitHub suggest (hey man, it's so complex to use that you need to study a book in order to understand it and learn that divine "git workflow")
Of course there are two parts here:
Maybe SVN but itself is also ugly and I just enjoyed the easy of use of TortoiseSVN, but the fact is that I've been trying to use the GUI that they provided for windows, so maybe the problem isn't Git by itself and it's just GitHub for Windows.
The first (ugly) surprise is that instead of using the common controls and theme of my windows, it's using that flat look of windows 8 in "modern" mode.
Hint: If I'm using certain version of windows and I'm using a theme (just the classic one, nothing fancy), it might be because I enjoy it and I don't want to use Windows8 even if you pay me.
Everything is flat, there's no real hint about which elements are simple text and which ones are active ones that can be used for something.
The colors are white background, grey text and some highlights; it's sad, no distinctive elements to clearly show the different zones and possibilities that the UI offers.
Or maybe the problem is that so few things are possible, I'm always thinking that I must be doing something wrong because I can't understand how to change things.
Last week I created a repository for a new little project. Sounds easy right?
It showed up with a lock icon (instead of the book like the other ones) and despite my efforts I haven't managed to make it show the readme.md in local.
A long time ago I published a basic example showing the skeleton of how an upload script for CKEditor should look like. Since the release of the imagepaste plugin I've received some requests to fill and explain all the details, and some people buy that plugin even before testing the free version and then they find that they don't know how to upload files.
So now I'm publishing that script with the missing part, a simple call to
move_uploaded_file($_FILES["upload"]["tmp_name"], $basePath . $name);
I've adjusted it a little so that the path to the upload folder and matching URL can be configured more easily, but it's still a very basic script. It doesn't perform any check on the uploaded files, you must take care of integrating it with your CMS (after all, that's something impossible for me to know how you're doing that part) and verify that the upload folder has the right permissions.
You can use it with CKEditor to enable the Upload tab in the Image and Link dialogs, and that way the user will be able to upload files (but not browse the existing ones). If you add the free version of the image paste plugin, then the user will be able to paste images from their clipboard into CKEditor, and the content will not be that huge mess of base64 encoded data, but a simple URL to your server with all the performance benefits that it means.
At this point, you can opt to buy the full version of the image paste plugin if you want to provide even more options for the users:
- Drag and drop from the computer for every modern browser
- Pasting images from the clipboard in Chrome (besides the basic pasting available in the free version for Firefox)
- Copy and paste a file from your computer in Firefox
- Two buttons to quickly upload a file or image for every browser. There's no way to make uploads simpler than this, other plugins still require the user to open a dialog, click browse, upload, etc... that's a huge mess and a big waste of time. Here you click and the file picker launches automatically, select the file that you want and then it's uploaded and inserted in a single step.
Back to this upload script:
I've published it at GitHub so if other people wants to provide some suggestions for better compatibility or they take the time to improve it I will try to keep it updated, but I don't really want to spend too much time on it because if by a miracle it would be a success (in fact, I wouldn't like to see too many people using this because it's really basic) then attackers would love to exploit all the weakness that it has. Please, read the documentation, all the comments and try your best to secure it as much as you can. If you download it from GitHub, you can remove those extra .git* files that it has added automatically as well as the readme and the docs when you have read them.
You only need the upload.php file and to improve a little the security by obscurity you can opt to rename it anyway that you want, just put that same name and path in the configuration of your CKEditor.
The best option would be for you to take this script as a full example and then add your customizations, your own security options, make it unique so that a simple bot can't find and use it.
This is the link to download the current version from GitHub.
Now that it's summer time and I have some free time I decided to take a look at the beta of IE 11 to check how well does it work with some of my CKEditor plugins.
The first thing to remember is that CKEditor itself isn't compatible with IE11 and if we are to base our guesses on previous history I don't think that they will really work on it until sometime after IE11 has been released and people start complaining seriously, but I prefer to look ahead and try to work now instead of having to rush later.
So I've focused my tests on the image paste plugin, specially because in their release notes they talk about the new ability to use the .files property in the clipboardData object as well as the base64 encoding performed by default (like Firefox).
Well, using base64 on the image is a nice little trick, but I think that it's really a bad option in the real world. I guess that if any performance guru would have been consulted they would have answered that this feature shouldn't have been added (at least not by default).
Testing IE nowadays it's much simpler than it used to be. Microsoft provides virtual machines ready to use with almost any combination of supported OS and IE as well as the virtual machine environment (at least I didn't miss anything). So after downloading the Win 8.1 preview and "installing" it, I was ready to test.
Ok, almost ready to test: no network connection. It turns out that the default network configuration is set up to work as a private network with the host. After changing it to my default bridged preference I had Internet again :-)
So I launched my demo with CKEditor 3 and... nothing. Nothing is pasted there. Not only the image is ignored from the clipboard, also any text can't be pasted. As the argument will be that CKEditor is "old" and it's not supported I moved then to the CKEditor 4 demo to check it, and yes, now it "works". The image is pasted but as base64 data. That means that neither my direct paste listener on the element is working to detect the files and the default CKEditor paste event is also failing to provide me the base64 so I could try to use the same Firefox trick.
A little debugging shows that the CKEditor paste isn't fired at all and my listener on the editable element works, but the event doesn't seem to include the .files property. Debugging that info isn't as easy as it seems, despite the claim that after pressing F12 the debugger is ready isn't true for me, breakpoints doesn't work, the files show up as a huge list of "plugin.js" files...
At this point a very important suggestion for Microsoft: Show folder names in the pane with the loaded scripts, having around 30 or 50 files named "plugin.js" and no way to tell them apart is not fun.
In order to debug correctly I restart the machine (maybe it's usual for people to start debuggin on the first IE & OS launch, but I don't see any other reason to fire up that VM, Win8.1 still tries to force "metro" too much so it doesn't fit my tastes) and now things seem to work a little better. To find the place for my breakpoints I search the files for the some function name specific to that file and that way I can avoid the mess of "plugin.js" .
And now looking at their documentation I notice something interesting, instead of using a clipboardData object on the paste event data, they are using one one the window object (like old IEs), so now it's obvious why this detection wasn't working. I add a workaround to try to use it and now when I copy an image from Paint I can detect and upload it like it happens in Firefox and Chrome. Due to the way that they have written their documentation I think that this feature isn't complete, after all I haven't managed to get an event if I try to paste a file from the desktop (like Firefox does) and if I create a document in WordPad with an image only the text is copied and no image is inserted, although given the way that they have chosen to write the documentation I think that some other feature should be available in the final version besides pasting a simple image from the clipboard.
After noticing that issue and given that this is a problem that shouldn't exist (providing files in the clipboard data is a feature that currently only Firefox and Chrome support, so if Microsoft want existing code to work they should use exactly the same APIs and objects) I filed a bug in their tracker to check if they care about compatibility or not. To my surprise this morning I had a reply, stating that obviously they can't reproduce the bug and the best course of action is to close it. I've replied providing the test file as a live url instead of an attachment (they didn't allow .html attachments so finding a .txt page might have been too hard for them). Of course this probably means the end of my efforts to report any issue in IE11 to Microsoft, if they don't care about something so simple and basic according to their goal of treating IE11 as any other modern browser then I'm not gonna waste my time trying to explain more complex things.
Back to my testings on IE11 and image paste, I noticed that dropping a file doesn't upload it and instead tries to open it as if I hadn't added any drop listeners. In this case the problem is that CKEditor at the moment detects IE11 as a gecko, but in this case Microsoft is aligning their behavior to old Chrome and requires that the drag over event is prevented. This is the kind of bug that Microsoft should also fix, but I don't have hope on them so I just add a line to detect "IEmodern" and it works (yes, this is the kind of bug/browser behavior that can't be changed based on feature detection but instead it relies on hardcoding which are the buggy browsers)
So now I tested again the CKEditor 3 demo and everything related to my plugin works. Images can be pasted, dragged, uploaded with the toolbar button... CKEditor itself has its own set of bugs (like not being able to paste at all in 3.6 or that the paste event isn't fired in 4.2) but that's outside my control. I could try to debug and provide patches if I really cared, but given that some of the patches that I provided for CKEditor have been ignored for years I prefer to not waste the time that way.
BTW, soon I might replace all the CKEditor 4 demos with the "Moono color" skin, being forced to see those black and white icons is really depressing.