Bug 87154 – Make the files attribute of HTMLInputElement writable

• Component: New Bugs
• Status: RESOLVED FIXED

[Nico Weber]

Make the files attribute of HTMLInputElement writable

[Nico Weber]

Created attachment 143343 [details]
Patch

[Nico Weber]

Comment on attachment 143343 [details]
Patch

still needs tests

[Nico Weber]

Created attachment 143374 [details]
Patch

[Adam Barth]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

> Source/WebCore/html/FileInputType.cpp:224
> +void FileInputType::setFiles(PassRefPtr<FileList> files)

Did you just move this function, or did you change something about it?

[Nico Weber]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

>> Source/WebCore/html/FileInputType.cpp:224
>> +void FileInputType::setFiles(PassRefPtr<FileList> files)
> 
> Did you just move this function, or did you change something about it?

I just moved it. See

"""
(WebCore::FileInputType::setFiles):
    Pure code move.
"""

in the ChangeLog. Is it better to leave comments on the diff too?

[Darin Adler]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

>> Source/WebCore/html/FileInputType.cpp:224
>> +void FileInputType::setFiles(PassRefPtr<FileList> files)
> 
> Did you just move this function, or did you change something about it?

Let me make a stronger statement. There is no need to move this function. Please don’t move it.

> Source/WebCore/html/HTMLInputElement.cpp:1226
> +    if (files)
> +        m_inputType->setFiles(files);

Why a null check here but not in FileInputType's setFiles function?

[Build Bot]

Comment on attachment 143374 [details]
Patch

Attachment 143374 [details] did not pass win-ews (win):
Output: http://queues.webkit.org/results/12761174

[Adam Barth]

> in the ChangeLog. Is it better to leave comments on the diff too?

Nope, the ChangeLog is the right place.  I just failed to read it.  :)

[Nico Weber]

Created attachment 143382 [details]
Patch

[Nico Weber]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

>>>> Source/WebCore/html/FileInputType.cpp:224
>>>> +void FileInputType::setFiles(PassRefPtr<FileList> files)
>>> 
>>> Did you just move this function, or did you change something about it?
>> 
>> I just moved it. See
>> 
>> """
>> (WebCore::FileInputType::setFiles):
>>     Pure code move.
>> """
>> 
>> in the ChangeLog. Is it better to leave comments on the diff too?
> 
> Let me make a stronger statement. There is no need to move this function. Please don’t move it.

Undone.

>> Source/WebCore/html/HTMLInputElement.cpp:1226
>> +        m_inputType->setFiles(files);
> 
> Why a null check here but not in FileInputType's setFiles function?

The if here protects the call to FileInputType's setFiles() function, so if that's called the pointer has to be NULL. If you prefer, I can check in both places.

[Nico Weber]

Created attachment 143383 [details]
Patch

[Ojan Vafai]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

>>> Source/WebCore/html/HTMLInputElement.cpp:1226
>>> +        m_inputType->setFiles(files);
>> 
>> Why a null check here but not in FileInputType's setFiles function?
> 
> The if here protects the call to FileInputType's setFiles() function, so if that's called the pointer has to be NULL. If you prefer, I can check in both places.

I think Darin's point was to put it in FileInputType's setFiles (and not here) because that's where the pointer is actually used.

[Nico Weber]

Comment on attachment 143374 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143374&action=review

>>>> Source/WebCore/html/HTMLInputElement.cpp:1226
>>>> +        m_inputType->setFiles(files);
>>> 
>>> Why a null check here but not in FileInputType's setFiles function?
>> 
>> The if here protects the call to FileInputType's setFiles() function, so if that's called the pointer has to be NULL. If you prefer, I can check in both places.
> 
> I think Darin's point was to put it in FileInputType's setFiles (and not here) because that's where the pointer is actually used.

Done.

(My reasoning for putting it here: Otherwise, all the InputType subclasses that want to implement setFiles() need to NULL-check (admittedly, at the moment that's only FileInputType). If the if is here, it's only in one place. Also, before this patch, FileInputType::setFiles() could never be called with a NULL pointer and I felt maintaining that would be good.)

[Nico Weber]

Created attachment 143388 [details]
Patch

[Ojan Vafai]

Comment on attachment 143388 [details]
Patch

Patch looks good. If you don't mind, lets allow the whatwg thread a couple days to give time for other browser vendors to voice concerns.

http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2012-May/036140.html

[Kent Tamura]

Comment on attachment 143388 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=143388&action=review

> Source/WebCore/html/HTMLInputElement.idl:35
>          readonly attribute HTMLFormElement form;
> -        readonly attribute FileList files;
> +        attribute FileList files;
>          attribute [Reflect, URL] DOMString formAction;

Why?

[Nico Weber]

(In reply to comment #16)
> (From update of attachment 143388 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=143388&action=review
> 
> > Source/WebCore/html/HTMLInputElement.idl:35
> >          readonly attribute HTMLFormElement form;
> > -        readonly attribute FileList files;
> > +        attribute FileList files;
> >          attribute [Reflect, URL] DOMString formAction;
> 
> Why?

From the linked whatwg thread (http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2012-May/036140.html):

"""Making the attribute writable would allow setting the files property
of an input element to dataTransfer.files from a drop handler. For
example, I would like to use this to create a larger drop-target for a
file input. Here's one request for this functionality:
http://stackoverflow.com/questions/8006715/drag-drop-files-into-standard-html-file-input"""

[Kent Tamura]

(In reply to comment #17)
> (In reply to comment #16)
> > (From update of attachment 143388 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=143388&action=review
> > 
> > > Source/WebCore/html/HTMLInputElement.idl:35
> > >          readonly attribute HTMLFormElement form;
> > > -        readonly attribute FileList files;
> > > +        attribute FileList files;
> > >          attribute [Reflect, URL] DOMString formAction;
> > 
> > Why?
> 
> From the linked whatwg thread (http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2012-May/036140.html):
> 
> """Making the attribute writable would allow setting the files property
> of an input element to dataTransfer.files from a drop handler. For
> example, I would like to use this to create a larger drop-target for a
> file input. Here's one request for this functionality:
> http://stackoverflow.com/questions/8006715/drag-drop-files-into-standard-html-file-input"""

I think we don't have consensus of the spec change yet.  It's too early to implement it.

[Nico Weber]

Yes, see Ojan's "If you don't mind, lets allow the whatwg thread a couple days to give time for other browser vendors to voice concerns." in comment 15.

[Build Bot]

Comment on attachment 143388 [details]
Patch

Attachment 143388 [details] did not pass win-ews (win):
Output: http://queues.webkit.org/results/12761192

[Build Bot]

Comment on attachment 143388 [details]
Patch

Attachment 143388 [details] did not pass mac-ews (mac):
Output: http://queues.webkit.org/results/12768019

[Ojan Vafai]

Comment on attachment 143388 [details]
Patch

Lets hold off on landing this till the whatwg discussion settles on what the best thing to expose is.

[Adam Barth]

Sounds like we should move forward here.  The response on WhatWG has been basically "yes, let's do this and more."  We can start with this and then do more later.

[Adam Barth]

Nico, would you be willing to update the patch so that the mac and win bubbles will be green?

[Nico Weber]

Created attachment 144680 [details]
Patch

[Nico Weber]

(In reply to comment #24)
> Nico, would you be willing to update the patch so that the mac and win bubbles will be green?

I'm giving it a try. If they do become green, someone familiar with the ObjC bindings generator should look at the change to PublicDOMInterfaces.h.

[WebKit Review Bot]

Please wait for approval from [e-mail] (or another member of the Apple Safari Team) before submitting because this patch contains changes to the Apple Mac WebKit.framework public API.

[Timothy Hatcher]

Comment on attachment 144680 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=144680&action=review

> Source/WebCore/bindings/objc/PublicDOMInterfaces.h:629
> +#if WEBKIT_VERSION_MIN_REQUIRED >= WEBKIT_VERSION_LATEST
> +@property(retain) DOMFileList *files;
> +#else
>  @property(readonly, retain) DOMFileList *files AVAILABLE_IN_WEBKIT_VERSION_4_0;
> +#endif

#ifdefs in this file are not correct. This file is parsed by a build script and not used by a compiler. In this case it is file to just drop the readonly and keep the AVAILABLE_IN_WEBKIT_VERSION_4_0.

[Timothy Hatcher]

"it is file" => "it is fine"

[Nico Weber]

Created attachment 144681 [details]
Patch

[Build Bot]

Comment on attachment 144681 [details]
Patch

Attachment 144681 [details] did not pass mac-ews (mac):
Output: http://queues.webkit.org/results/12840783

[Nico Weber]

(In reply to comment #31)
> (From update of attachment 144681 [details])
> Attachment 144681 [details] did not pass mac-ews (mac):
> Output: http://queues.webkit.org/results/12840783

The error is:

make: *** [JSAudioPannerNode.h] Segmentation fault: 11

I'm not sure if that's related – is there a way to rerun an ews bot?

[Build Bot]

Comment on attachment 144681 [details]
Patch

Attachment 144681 [details] did not pass win-ews (win):
Output: http://queues.webkit.org/results/12852643

[Nico Weber]

(In reply to comment #33)
> (From update of attachment 144681 [details])
> Attachment 144681 [details] did not pass win-ews (win):
> Output: http://queues.webkit.org/results/12852643

win ews error:

34>c:\cygwin\home\buildbot\WebKit\Tools\TestWebKitAPI\config.h(47) : fatal error C1083: Cannot open include file: 'WebKit2/WebKit2.h': No such file or directory

Looks unrelated as well :-/

[Nico Weber]

Created attachment 144840 [details]
Patch

[Build Bot]

Comment on attachment 144840 [details]
Patch

Attachment 144840 [details] did not pass win-ews (win):
Output: http://queues.webkit.org/results/12840976

[Nico Weber]

Created attachment 144856 [details]
Patch

[Nico Weber]

Now with green mac and win bubbles. (The mac bubble went green when I reuploaded the patch without any changes; for the win bubble I needed to add a `#include "FileList.h"` to FileType.cpp -- else cl.exe would complain about PassRefPtr<FileList>.)

[Adam Barth]

Comment on attachment 144856 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=144856&action=review

> Source/WebCore/bindings/objc/PublicDOMInterfaces.h:625
> -@property(readonly, retain) DOMFileList *files AVAILABLE_IN_WEBKIT_VERSION_4_0;
> +@property(retain) DOMFileList *files AVAILABLE_IN_WEBKIT_VERSION_4_0;

I believe this is backwards compatible, so this should be fine.

[Nico Weber]

(In reply to comment #39)
> (From update of attachment 144856 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=144856&action=review
> 
> > Source/WebCore/bindings/objc/PublicDOMInterfaces.h:625
> > -@property(readonly, retain) DOMFileList *files AVAILABLE_IN_WEBKIT_VERSION_4_0;
> > +@property(retain) DOMFileList *files AVAILABLE_IN_WEBKIT_VERSION_4_0;
> 
> I believe this is backwards compatible, so this should be fine.

It's also what TImothy told me to do in comment 28 :-)

[Adam Barth]

Comment on attachment 144856 [details]
Patch

Looks like Timothy Hatcher has already approved the PublicDOMInterfaces.h part of this change.  Thanks!

[Adam Barth]

> It's also what TImothy told me to do in comment 28 :-)

Yeah, I'm just a little slow.  :)

[WebKit Review Bot]

Comment on attachment 144856 [details]
Patch

Clearing flags on attachment: 144856

Committed r118993: <http://trac.webkit.org/changeset/118993>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Ian 'Hixie' Hickson]

Is this documented anywhere?

[Kent Tamura]

(In reply to comment #45)
> Is this documented anywhere?

http://www.chromium.org/developers/web-platform-status/forms
| files property is writable. [M21]

Safari in iOS7 should have this behavior. But I don't know if it's documented.