Workaround for Edge not supporting custom MIME types in `dataTransfer.setData` #968

f1ames · 2017-09-27T14:56:55Z

What is the purpose of this pull request?

Task

Does your PR contain necessary tests?

All patches which change the editor code must include tests. You can always read more
on PR testing,
how to set the testing environment and
how to create tests
in the official CKEditor documentation.

This PR contains

Unit tests
Manual tests

What changes did you make?

Closes #962, #1026.

f1ames · 2017-09-27T14:58:18Z

tests/plugins/pastefromword/parsestyles.html

@@ -39,7 +39,7 @@
 	mso-style-qformat:yes;
 	mso-style-parent:"";
 	margin:0cm;
-	margin-bottom:.0001pt;
+	margin-bottom:.001pt;


Edge rounds .0001 to 0 when parsing stylesheet so I changed it to .001 which works fine.

mlewand · 2017-10-09T14:17:32Z

tests/_benderjs/ckeditor/static/tools.js

+					}
+
+					if ( CKEDITOR.env.ie && CKEDITOR.env.version >= 16 &&
+						CKEDITOR.tools.indexOf( [ 'Text', 'URL', 'text/plain', 'text/html', 'application/xml' ], type ) === -1 ) {


A short comment explanation, would be useful here.

Also, I believe it's valid only for Edge, isn't it?

mlewand · 2017-10-09T14:36:56Z

plugins/clipboard/plugin.js

@@ -2013,6 +2021,11 @@
 			var nativeDataTransfer = evt.data.$ ? evt.data.$.dataTransfer : null,
 				dataTransfer = new this.dataTransfer( nativeDataTransfer, sourceEditor );

+			// Set dataTransfer.id only for 'dragstart' event (so for events initializing dataTransfer inside editor) (#962).
+			if ( evt.name === 'dragstart' && clipboardIdDataType !== 'Text' ) {
+				dataTransfer.setData( clipboardIdDataType, dataTransfer.id );


I think it could be extracted as a private method, like dataTransfer._setId() or sth.

With this change this test could be simplified: https://github.com/ckeditor/ckeditor-dev/blob/ea41a52/tests/plugins/clipboard/datatransfer.js#L39

Introduced dataTransfer.setId method. As it is called from external place (not inside CKEDITOR.plugins.clipboard.dataTransfer) it shouldn't be private IMHO.

mlewand · 2017-10-09T16:14:12Z

tests/plugins/clipboard/datatransfer.js

@@ -482,7 +492,10 @@ bender.test( {
 			// Emulate native clipboard.
 			nativeData = bender.tools.mockNativeDataTransfer();

-		if ( isCustomDataTypesSupported ) {
+		// This test uses native (mocked) `setData` which does not applies fallback


I find this comment is vague, first "native (mocked)" is a no go - is it native or a mock?

mlewand · 2017-10-09T16:15:49Z

tests/plugins/clipboard/datatransfer.js

@@ -553,7 +566,8 @@ bender.test( {

 	// http://dev.ckeditor.com/ticket/12961
 	'test file in items': function() {
-		if ( CKEDITOR.env.ie && CKEDITOR.env.version < 10 ) {
+		// DataTransfer.items is not supported in IE/EDGE so there is no reason to test it.


Important: actually AFAIR it works for drag events for IE11 (presumably also for Edge) so there is a good reason to keep that enabled.

mlewand · 2017-10-09T16:16:12Z

tests/plugins/clipboard/datatransfer.js

@@ -642,7 +656,8 @@ bender.test( {

 	// http://dev.ckeditor.com/ticket/12961
 	'test file in items with cache': function() {
-		if ( CKEDITOR.env.ie && CKEDITOR.env.version < 10 ) {
+		// DataTransfer.items is not supported in IE/EDGE so there is no reason to test it.
+		if ( CKEDITOR.env.ie ) {


Same as above.

mlewand · 2017-10-09T16:56:23Z

plugins/clipboard/plugin.js

@@ -2296,6 +2317,21 @@
 				} catch ( e ) {}
 			}

+			// If we are getting the same type which may store custom data we need to extract it (removing custom data comment).
+			// Or if we are getting different type and it is empty we need to check inside data comment if it is stored there.
+			if ( !isEmpty( data ) && type === this.customDataFallbackType ) {


With this construction every modern browser will take a performance hit when calling dt.getData( 'text/html' ). Ideally we want to do this only for browsers that really need it (Edge 16).

mlewand · 2017-10-09T16:59:49Z

plugins/clipboard/plugin.js

+				content = this._extractDataComment( data );
+				data = content.content;
+
+			} else if ( isEmpty( data ) ) {


Couldn't we do this part in this catch statement https://github.com/ckeditor/ckeditor-dev/blob/ea41a52/plugins/clipboard/plugin.js#L2326 ?

AFAIR Edge throws an exception when calling getData() on non-whitelisted type. In that case we could check for exception type/message, and if it's an exception denying access to a requested dataTransfer type, then attempt to extract data comments in the fallbackType.

The point is to execute this (possibly intense) operation as little as possible. This is a clipboard, so these kind of optimizations are important.

Unfortunately, Edge only throws error on setData for not supported mime types, getData just returns empty string. However, after refactoring from #968 (comment) this issue will no longer be a problem (to some extent, see #1026).

mlewand · 2017-10-09T17:30:10Z

plugins/clipboard/plugin.js

@@ -114,6 +114,8 @@
 'use strict';


General note:

I'd be in favor to extract the whole logic to a separate object. Say

var fallbackClipboard: { setData: (){}, getData: (){}, _setCustomData: (){}, _getCustomData: (){}, _extractDataComment: (){}, _applyDataComment: (){} }

And then original setData could be restored to previous, cleaner, form - and not expose reader for lengthy internal logic.

I see it like sth like this:

try { this.$.setData( type, value ); } catch ( e ) { // Just guessing how an error might be recognized. if ( e instanceof PermissionError ) { fallbackClipboard.setData( this, type, value ); } }

Note that fallbackClipboard.setData() and fallbackClipboard.getData() would need to feature an internal try/catch structure.

Instead of:

// Extract custom data if used type is the same one which is used for storing it. // Then new value is added to a custom data so it will not be overwritten entirely. if ( type === this.customDataFallbackType ) { var content = null; try { content = this._extractDataComment( this.$.getData( this.customDataFallbackType ) ); } catch ( e ) {} if ( content && content.data ) { value = this._applyDataComment( value, content.data ); } this._setCustomData( type, value ); } else { this._setCustomData( type, value, CKEDITOR.env.ie && CKEDITOR.env.version >= 16 ); }

Also I believe it would be very beneficial to cache customData rather than fetch it every time. It's very easy to know when the cache should be invalidated.

And again, since we implement this fallback in a separate type all of this will be cleaner. Though I'm OK with taking care of cache in a follow-up ticket.

👍 Created issue for caching #1026.

mlewand · 2017-10-09T17:49:50Z

tests/plugins/clipboard/manual/datatransferfallback.html

@@ -0,0 +1,211 @@
+<head>


For the love of god, please simplify this manual test. 😄 It's enough for us to have this monster in generic tests.

Reduce toolbars, editors content just to contain bare essential.

Ask yourself what is important in this test, and whether you provide correct tools to check it. Imagine you see it for the very first time, and you know nothing.

Yeah, I also was overwhelmed when seeing this generic test for the first time. And then somehow I duplicated it...
I simplified the test to contain only the most important information/scenario.

mlewand · 2017-10-09T18:09:47Z

tests/plugins/pastefromword/parsestyles.js

@@ -12,6 +12,14 @@
 		bender.tools.testInputOut( name, function( styles, expected ) {
 			var tested = CKEDITOR.plugins.pastefromword.styles.inliner.parse( styles );

+			// In CKEDITOR.plugins.pastefromword.styles.inliner.parse#createIsolatedStylesheet function
+			// somehow Edge camelcases the selectors so we need to lowercase it (#962).


This should be troubleshooted, and reported to upstream, sounds like a serious issue. Might be coming from here: https://github.com/ckeditor/ckeditor-dev/blob/ea41a52/plugins/pastefromword/filter/default.js#L810-L820

In which case it would be crucial problem for our styles inlining.

Luckily (😌) I smell something fishy is happening in our codebase 🙂 See the expected value in: https://github.com/ckeditor/ckeditor-dev/blob/ea41a52/tests/plugins/pastefromword/parsestyles.html#L7-L12

It should be camelcased so Edge provides a correct value, the problem is why expected value gets lowercased?

After I have logged it it turns out that chrome also puts the expected value as lowercase, which doesn't make sense. I'll leave it here for you.

I extracted it to a separate issue as it seems to be quite odd behaviour #1042.

…ransfer'.

mlewand

I have created a manual test that fails on latest edge: http://tests.ckeditor.test:1030/tests/plugins/clipboard/manual/customtypes

If you check Edge, it puts only last of custom data types, and discards text/html

While other browsers works correctly:

I did some debugging but it took some solid amount of time, without getting the source of the issue.

My initial assumption was that Edge does not return correct results when (native) clipboardData.getData() is called right after setting the data, however after running listener like that:

evt.editor.editable().on( 'copy', function( ckEvt ) {
	var evt = ckEvt.data.$,
		nativeDataTransfer = evt.clipboardData,
		ckDataTransfer = new CKEDITOR.plugins.clipboard.dataTransfer( nativeDataTransfer );

	ckDataTransfer.setData( 'text/plain', 'plain value' );
	ckDataTransfer.setData( 'text/html', 'html value' );
	ckDataTransfer.setData( 'foo-bar', 'foo-bar value' );
	ckDataTransfer.setData( 'baz', 'baz value' );

	console.log( 'plain', ckDataTransfer.getData( 'text/plain' ) );
	console.log( 'html', ckDataTransfer.getData( 'text/html' ) );

	// Prevent DOM event, so that clipboard gets overriden.
	evt.preventDefault();
}, null, null, 100000 );

It works well, so the issue is something other.

mlewand · 2017-10-24T13:04:39Z

plugins/clipboard/plugin.js

+		 * @param {String} value
+		 */
+		setData: function( type, value ) {
+			// Extract custom data if used type is the same one which is used for storing it.


I think this code is just a bit misleading: it's not only extracting, but also appending this data.

How about something like that?

// In case of setting type that stores custom data, make sure that current custom data gets applied.

mlewand · 2017-10-24T13:39:36Z

And the question is also why unit tests didn't catch it 🤔

f1ames · 2017-10-25T08:55:34Z

The issue is caused by the upstream bug (https://developer.microsoft.com/en-us/microsoft-edge/platform/issues/14110451/). It is not possible to retrieve the data (via nativeDataTransfer.getData) in the same listener (synchronously) in which it was set.

That' the reason it looks like setting custom types overwrites previous values. The mechanism check if something was already checked in merges the data. As the browse incorrectly returns empty data, the new value does not contain previously set data.

And the question is also why unit tests didn't catch it

Unit tests use data transfer mock so native issues may not be visible there.

The solution as we discussed with @mlewand will be to implement cache mechanism for CKEDITOR.plugins.clipboard.fallbackDataTransfer. It should utilize already existing cache mechanism from CKEDITOR.plugins.clipboard.dataTransfer.

msamsel · 2017-10-31T13:01:07Z

Extracted #1047.

I've made fix for that, which runs test in separate editors.

mlewand

The problem is clipboard is getting bigger and bigger. It has just passed 3k LoC milestone. 😰

I was thinking about something much simpler, instead of adding (backward incompatible) full featured type for caching, why not use "old" and simple cache object? The only thing you need on top of it is a list of custom types that should be handled by _fallbackDataTransfer.

How to do that?

Just keep a simple array, say CKEDITOR.plugins.clipboard.fallbackDataTransfer.customTypes (it make sense to make it a static prop). It's members are collected when:

With that we'd have:

100% backward copatibility
substantially less LoC
simplified implementation

Additionally:

It would be best to stick with the existing class structure, and put cache instance in dataTransfer._.
- Actually WDYT, does it make sense for us to move fallbackDataTransfer there too?

mlewand · 2017-11-06T10:57:05Z

tests/plugins/clipboard/manual/customtypes.html

+					if ( CKEDITOR.tools.array.indexOf( types, 'text/html' ) != -1 ) {
+						// Native 'getData' needs to be used here as 'ckDataTransfer.getData'
+						// strips special comment which we would like to show here too.
+						output += getRow( 'Native HTML data', ckDataTransfer.$.getData( 'text/html', true ), 'raw' );


I think that in fact special comment should be included when calling ckDataTransfer.getData( 'text/html', true ). By passing getData() getNative parameter, you say that you really mean you want to get all the meat - and this is what you should get.

So that being said fallback should respect that and return raw text/html. This parameter was created deliberately for text/html (though it's not properly mentioned in the API docs 😒 #1121).

Makes sense 👍 , I wasn't fully sure how it should work with fallback html.

Btw. should getNative work only for text/html? In Edge if you call getData( 'cke/id', true ) if we assume getNative should really return what natively sits in dataTransfer you will get empty string.

The other issue with getNative flag is that, it is used inside cacheData:
https://github.com/ckeditor/ckeditor-dev/blob/96bf4037cb3ff812fc9b993bce4a304a884f361a/plugins/clipboard/plugin.js#L2450

so we would like to cache "native value" for all browsers, but for Edge the cache should contain proper type : value pairs without special comment so it shouldn't basically use getNative (fallbackDataTransfer.isRequired() could be use here but it will create another "logic branch", increasing the complexity of already complex plugin 😞 ).

Btw. should getNative work only for text/html? In Edge if you call getData( 'cke/id', true ) if we assume getNative should really return what natively sits in dataTransfer you will get empty string.

This is exactly how it should work, I can see that you already implemented it that way. I have corrected this in 3375736.

mlewand · 2017-11-06T11:28:57Z

plugins/clipboard/plugin.js

+				value = null;
+
+			// If we are getting the same type which may store custom data we need to extract content only.
+			if ( type === this._customDataFallbackType ) {


What about plain text? AFAICT in this construction it will prioritize looking for plain text in the dataComment.

Yes, it will. But as we don't have hardcoded list of supported types, first checking dataComment is required. Remember that accessing unsupported type in Edge returns empty string (no exception so it can't be distinguished properly). As plain/text won't be there it will simply fallback to regular getData.

Make sense 👍

mlewand · 2017-11-06T11:30:00Z

plugins/clipboard/plugin.js

+			if ( type === this._customDataFallbackType ) {
+				value = dataComment.content;
+
+			// If we are getting different type we need to check inside data comment if it is stored there.


This comment should be moved to a proper line, which is before if ( dataComment.data && dataComment.data[ type ] ) {.

mlewand · 2017-11-06T11:30:25Z

plugins/clipboard/plugin.js

+				if ( dataComment.data && dataComment.data[ type ] ) {
+					value = dataComment.data[ type ];
+
+				// And then fallback to regular `getData`.


Similar as above, it would make it more readable to move this comment to else block.

Without this change values were a bit different than expected pattern.

mlewand · 2017-11-06T12:35:45Z

I have pushed small change in the manual test in 0aac1fe - without it the values in table were not 100% following the pattern. (it has shown "plain value" instead of "Text value").

f1ames · 2017-11-06T13:24:06Z

I was thinking about something much simpler, instead of adding (backward incompatible) full featured type for caching, why not use "old" and simple cache object? The only thing you need on top of it is a list of custom types that should be handled by _fallbackDataTransfer.

Makes sense, I extracted it to make it more readable as getData/setData are already a little complicated, but we can try with simple version.

Just keep a simple array, say CKEDITOR.plugins.clipboard.fallbackDataTransfer.customTypes (it make sense to make it a static prop). It's members are collected when:

setData fails, due to a custom type,

getData (after adding a this._isUnsupportedMimeTypeError check)

We can only mark it on setData as getData does not throw an exception when accesing not supported type (returning empty string). But, it is the same as it works now.

It would be best to stick with the existing class structure, and put cache instance in dataTransfer._.
Actually WDYT, does it make sense for us to move fallbackDataTransfer there too?

Yes, I suppose both cache and fallbackDataTransfer should be in dataTransfer._. "namespace".

f1ames · 2017-11-07T10:29:47Z

I simplified the cache and added other small changes according to what is described above.

f1ames · 2017-11-07T16:20:12Z

I updated closes reference of this PR, as after introducing cache it also resolves #1026.

… native object.

mlewand

I must say, I like this view:

🙂

You have renamed dataTransfer._.data to dataTransfer._.cache. Let's make it dataTransfer._.data back, so any apps that relied on it won't break (though it was private/undocumented but there's no good reason to break it).

I feel these are just last touches here, and we're done with it.

mlewand · 2017-11-08T11:09:53Z

tests/plugins/clipboard/datatransfer.js

-			dataTransfer1a = new CKEDITOR.plugins.clipboard.dataTransfer( nativeData1 ),
-			dataTransfer1b = new CKEDITOR.plugins.clipboard.dataTransfer( nativeData1 ),
-			dataTransfer2 = new CKEDITOR.plugins.clipboard.dataTransfer( nativeData2 );
+			dataTransfer1a, dataTransfer1b, dataTransfer2;


Vars should be defined each in a separate line.

mlewand · 2017-11-08T11:31:59Z

plugins/clipboard/plugin.js

+	 * @param {Object} cache Cache object used on storing and retrieving data.
+	 * @param {Object} [nativeDataTransfer] A native data transfer object.
+	 */
+	CKEDITOR.plugins.clipboard.fallbackDataTransfer = function( cache, nativeDataTransfer ) {


Wouldn't it be better if this constructor simply take a (our) dataTransfer instance? Having that you could assign both cache and native data trasnfer.

I applied this fix, however then you have this._cache = dataTransfer._.data which means accessing private property outside of dataTransfer class, which shouldn't have place and is not a good practice. WDYT @mlewand, alternatively we could add a getter or leave it as it was?

mlewand · 2017-11-08T11:32:27Z

plugins/clipboard/plugin.js

+	 *
+	 * @private
+	 * @static
+	 * @property {Array}


You could doc the type even better, it's an array of strings, so {String[]}.

mlewand · 2017-11-08T11:44:35Z

plugins/clipboard/plugin.js

+		 * @param {String} value
+		 */
+		setData: function( type, value ) {
+			// In case of fallbackDataTransfer, cache does not reflect native data one-to-one. For example, having


Actually AFAICS this is no longer the case. Now cache reflects what's stored in given types (also cache does not contain custom data marker).

It is still the case, all data is cached in CKEDITOR.plugins.clipboard.dataTransfer#setData which doesn't have any knowledge how fallback works. So the cache contains proper type : value pairs no matter if it is supported by the browser or not.
And then native data transfer contains fallback specific types if needed.

AFAIR, that was one of the reasons for cache, to skip processing the special data comment every time when getData is called.

Ahh, sorry my bad, at first I looked at it as if 1st object was data mapping passed to .setData() and the second was our cache. While it is our cache -> browser native clipboard (which is actually documented).

I think we need to get other eyes on it to make sure it was only case for me. 🙂

mlewand · 2017-11-08T12:02:28Z

tests/plugins/clipboard/manual/customtypes.html

+					if ( CKEDITOR.tools.array.indexOf( types, 'text/html' ) != -1 ) {
+						// Native 'getData' needs to be used here as 'ckDataTransfer.getData'
+						// strips special comment which we would like to show here too.
+						output += getRow( 'Native HTML data', ckDataTransfer.$.getData( 'text/html', true ), 'raw' );


Btw. should getNative work only for text/html? In Edge if you call getData( 'cke/id', true ) if we assume getNative should really return what natively sits in dataTransfer you will get empty string.

This is exactly how it should work, I can see that you already implemented it that way. I have corrected this in 3375736.

mlewand · 2017-11-08T12:09:51Z

plugins/clipboard/plugin.js

+				value = null;
+
+			// If we are getting the same type which may store custom data we need to extract content only.
+			if ( type === this._customDataFallbackType ) {


Make sense 👍

…not support dataDransfer at all.

mlewand

And we made it, LGTM

mlewand · 2017-11-10T16:28:38Z

For the records, I've slightly messed the merge, because I rebased your branch onto latest major but then performed merge on GitHub without pushing.

To fix that I've quickly force-pushed t/468 branch back to where it was before the merge, rebased it onto latest major, and then merged rebased t/468-962 branch.

Now t/468 is fine. 🙂

Comandeer and others added 13 commits September 20, 2017 14:42

Switch on support for Clipboard API in Edge 15+.

bf67ea6

Updated tools for clipboard unit tests.

9d5d97f

Updated tests for dataTransfer.

b039172

Clipboard support from Edge 16.

36c4da1

Initial clipboard API support for Edge >= 16.

f39e02d

Edge workaround.

280d3c0

Unit tests.

cf26c37

Set id after setting other MIME types.

293fcc6

Edge version fix in tests.

f8c8e02

Manual test added.

2427a6f

Specs corrections.

76a1bb9

Fixed failing tests.

b431eca

Test fixes.

704a11f

f1ames commented Sep 27, 2017

View reviewed changes

f1ames added 3 commits September 28, 2017 14:09

Fix for 'OpenClipboard Failed' error.

2347c18

Set native dataTransfer id only on certain events.

95078d2

Tests: set native dataTransfer id only on certain events.

ea41a52

mlewand self-requested a review October 9, 2017 14:16

mlewand suggested changes Oct 9, 2017

View reviewed changes

Extracted all Edge logic to 'CKEDITOR.plugins.clipboard.fallbackDataT…

a5dba7f

…ransfer'.

f1ames mentioned this pull request Oct 10, 2017

Add caching to Clipboard API fallback implementation for Edge #1026

Closed

f1ames added 7 commits October 10, 2017 16:46

Introduced 'setId' method.

4eafeee

Docs fix.

596f2b2

Fallback data transfer improvements.

d6e8fcb

Fixed tests.

230e7ca

Handle empty 'dataTransfer' in 'isRequired' method.

0404325

Fixed failing test.

5d3abe1

Fix for 'dataTransfer.items'.

514f93c

f1ames mentioned this pull request Oct 12, 2017

Investigate Edge camelcasing selectors in 'CKEDITOR.plugins.pastefromword.styles.inliner.parse' #1042

Closed

Proper Edge workaround description.

cac60ad

Improved manual test.

5e52ce3

mlewand suggested changes Oct 24, 2017

View reviewed changes

f1ames added 3 commits October 26, 2017 16:54

Shared cache added to 'fallbackDataTransfer'.

e359599

Fix for 'getNative' param.

9b93460

Docs corrections.

96bf403

f1ames requested a review from mlewand October 26, 2017 16:15

mlewand suggested changes Nov 6, 2017

View reviewed changes

Corrected input data in manual test.

0aac1fe

Without this change values were a bit different than expected pattern.

f1ames added 2 commits November 7, 2017 11:17

Cache simplified and refactored.

a94c5bd

Tests: refactoring and adjusting to new cache structure.

9c0a25e

f1ames requested a review from mlewand November 7, 2017 10:32

mlewand added 2 commits November 8, 2017 12:37

Tests: corrected manual test not to leak undefined class.

3c621f4

Tests: use CKEditor dataTransfer API for getting raw HTML rather than…

3375736

… native object.

mlewand suggested changes Nov 8, 2017

View reviewed changes

Review fixes.

e61ac60

f1ames force-pushed the t/468-962 branch from 8621196 to e61ac60 Compare November 8, 2017 15:13

mlewand self-requested a review November 10, 2017 09:47

Tests: ignore datatransfer fallback manual test for browsers that do …

270fbda

…not support dataDransfer at all.

mlewand approved these changes Nov 10, 2017

View reviewed changes

mlewand merged commit 49281a9 into t/468 Nov 10, 2017

mlewand mentioned this pull request Nov 10, 2017

Fix for failing tests on Edge 16+ #1105

Closed

2 tasks

f1ames mentioned this pull request Nov 13, 2017

Support for Clipboard API in Edge #1153

Merged

2 tasks

Workaround for Edge not supporting custom MIME types in dataTransfer.setData #968

Workaround for Edge not supporting custom MIME types in dataTransfer.setData #968

Conversation

f1ames commented Sep 27, 2017 • edited Loading

What is the purpose of this pull request?

Does your PR contain necessary tests?

This PR contains

What changes did you make?

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlewand left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlewand commented Oct 24, 2017

f1ames commented Oct 25, 2017

msamsel commented Oct 31, 2017 • edited Loading

mlewand left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlewand Nov 6, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlewand commented Nov 6, 2017

f1ames commented Nov 6, 2017

f1ames commented Nov 7, 2017

f1ames commented Nov 7, 2017

mlewand left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mlewand left a comment

Choose a reason for hiding this comment

mlewand commented Nov 10, 2017

Workaround for Edge not supporting custom MIME types in `dataTransfer.setData` #968

Workaround for Edge not supporting custom MIME types in `dataTransfer.setData` #968

f1ames commented Sep 27, 2017 •

edited

Loading

msamsel commented Oct 31, 2017 •

edited

Loading

mlewand Nov 6, 2017 •

edited

Loading