Bug 113615 – Update all float attributes in HTMLMediaElement to double

• Status: RESOLVED
• Resolution: FIXED
• Component: Media

[Aaron Colwell]

HTMLMediaElement has several attributes declared as float that are actually specified as double in the HTML5 spec. These include currentTime, startTime, duration, defaultPlaybackRate, playbackRate, and volume. These should be updated to match the HTML5 spec.

[Aaron Colwell]

Created attachment 195821 [details]
Patch

[Eric Carlson]

Comment on attachment 195821 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=195821&action=review

This looks fine, but please wait until the bots have had their way with it before landing. Also, please file bugs against all ports that need to be updated, and also file one to remove the old versions once everyone has updated.

> Source/WebCore/ChangeLog:41
> +        (HTMLMediaElement):

These bogus entries generated by prepare-ChangeLog aren't helpful.

[Build Bot]

Comment on attachment 195821 [details]
Patch

Attachment 195821 [details] did not pass mac-ews (mac):
Output: http://webkit-commit-queue.appspot.com/results/17257710

[Build Bot]

Comment on attachment 195821 [details]
Patch

Attachment 195821 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-commit-queue.appspot.com/results/17311762

[Aaron Colwell]

Created attachment 195831 [details]
Remove ChangeLog cruft and fix Mac build

[Build Bot]

Comment on attachment 195831 [details]
Remove ChangeLog cruft and fix Mac build

Attachment 195831 [details] did not pass mac-ews (mac):
Output: http://webkit-commit-queue.appspot.com/results/17356171

[Build Bot]

Comment on attachment 195831 [details]
Remove ChangeLog cruft and fix Mac build

Attachment 195831 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-commit-queue.appspot.com/results/17255683

[Aaron Colwell]

Created attachment 195838 [details]
Another mac build fix.

[Build Bot]

Comment on attachment 195838 [details]
Another mac build fix.

Attachment 195838 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-commit-queue.appspot.com/results/17377035

[Build Bot]

Comment on attachment 195838 [details]
Another mac build fix.

Attachment 195838 [details] did not pass mac-ews (mac):
Output: http://webkit-commit-queue.appspot.com/results/17345825

[Aaron Colwell]

Created attachment 195845 [details]
Fix unused parameter error

[Build Bot]

Comment on attachment 195845 [details]
Fix unused parameter error

Attachment 195845 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-commit-queue.appspot.com/results/17317297

[Build Bot]

Comment on attachment 195845 [details]
Fix unused parameter error

Attachment 195845 [details] did not pass mac-ews (mac):
Output: http://webkit-commit-queue.appspot.com/results/17250676

[Aaron Colwell]

Eric, could you help me with the Mac build errors. I'm getting these symbol export errors.

Undefined symbols for architecture x86_64:
  "__ZN7WebCore16HTMLMediaElement14setCurrentTimeEfRi", referenced from:
     -exported_symbol[s_list] command line option
  "__ZN7WebCore16HTMLMediaElement9setVolumeEfRi", referenced from:
     -exported_symbol[s_list] command line option
  "__ZN7WebCore16HTMLMediaElement6rewindEf", referenced from:
     -exported_symbol[s_list] command line option
ld: symbol(s) not found for architecture x86_64

I'm assuming I have to change Source/WebCore/WebCore.order and/or Source/WebCore/WebCore.exp.in, but I don't have a Mac build environment so I have no idea what the new mangled signatures should be.

[Eric Carlson]

Created attachment 197271 [details]
Proposed patch

[Geoffrey Garen]

Comment on attachment 197271 [details]
Proposed patch

r=me

It would be nice to write a test that confirmed the additional precision. For example, you could seek to ".0000000000000000000000000000000000000000000001" and then ask for .currentTime and confirm that it's not 0.

[Eric Carlson]

(In reply to comment #16)
> (From update of attachment 197271 [details])
> r=me
> 
> It would be nice to write a test that confirmed the additional precision. For example, you could seek to ".0000000000000000000000000000000000000000000001" and then ask for .currentTime and confirm that it's not 0.

That won't work reliably because currentTime is the value returned by the media engine, and it will be clamped to whatever level of precision it supports.

[Radar WebKit Bug Importer]

<rdar://problem/13620078>

[Eric Carlson]

https://trac.webkit.org/r148099