Make cloning of File objects more useful.

Make cloning of File objects more useful.

Make the structured cloning of File objects complete, extending their
wire format to also include:

 * name
 * relative path
 * size
 * last modification date.

As this changes the format, bump the serialization version number up
to 4. The deserializer is backwards compatible, including for File
objects; verify this to be so through tests.

Files constructed via the File constructor weren't correctly cloned
as-was; added a Worker test to have some coverage.

R=dslomov@chromium.org,jochen@chromium.org
BUG=171905
TEST=fast/storage/serialized-script-value
TEST=fast/files/workers/worker-read-file-constructor-async

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=163956

[sof, 2013-12-15 21:39:42]

Please take a look when you next have a spare moment.

Bumping version of the wire format is unavoidable here, I think.

The deserialization code handles the current format & version just fine, but the
unknown (to me) is if there's a risk of v3-based deserializers being exposed to
this format (and perhaps needlessly fail as a result.)

[Dmitry Lomov (no reviews), 2013-12-16 09:28:55]

On 2013/12/15 21:39:42, sof wrote:
> Please take a look when you next have a spare moment.
> 
> Bumping version of the wire format is unavoidable here, I think.
> 
> The deserialization code handles the current format & version just fine, but
the
> unknown (to me) is if there's a risk of v3-based deserializers being exposed
to
> this format (and perhaps needlessly fail as a result.)

I think Rietveld ate your patch - I am only seeing 'chunk mismatch' for half of
the files.

Do we have a spec for structurally cloning Files? (I only see FileList here:
http://www.whatwg.org/specs/web-apps/current-work/multipage/common-dom-interf...)

Bumping wire format version is ok as long as you can still read the old
versions.

[sof, 2013-12-16 10:48:18]

On 2013/12/16 09:28:55, Dmitry Lomov (chromium) wrote:
> On 2013/12/15 21:39:42, sof wrote:
> > Please take a look when you next have a spare moment.
> > 
> > Bumping version of the wire format is unavoidable here, I think.
> > 
> > The deserialization code handles the current format & version just fine, but
> the
> > unknown (to me) is if there's a risk of v3-based deserializers being exposed
> to
> > this format (and perhaps needlessly fail as a result.)
> 
> I think Rietveld ate your patch - I am only seeing 'chunk mismatch' for half
of
> the files.
> 

Sorry about that, uploaded again.

> Do we have a spec for structurally cloning Files? (I only see FileList here:
>
http://www.whatwg.org/specs/web-apps/current-work/multipage/common-dom-interf...)
> 

File derives from Blob, so covered by its case.

> Bumping wire format version is ok as long as you can still read the old
> versions.

ok, the serialized-script-value test verifies that v3 formats are readable.

[Dmitry Lomov (no reviews), 2013-12-16 10:52:50]

lgtm

Thanks for the fix!

On 2013/12/16 10:48:18, sof wrote:
> On 2013/12/16 09:28:55, Dmitry Lomov (chromium) wrote:
> > On 2013/12/15 21:39:42, sof wrote:
> > > Please take a look when you next have a spare moment.
> > > 
> > > Bumping version of the wire format is unavoidable here, I think.
> > > 
> > > The deserialization code handles the current format & version just fine,
but
> > the
> > > unknown (to me) is if there's a risk of v3-based deserializers being
exposed
> > to
> > > this format (and perhaps needlessly fail as a result.)
> > 
> > I think Rietveld ate your patch - I am only seeing 'chunk mismatch' for half
> of
> > the files.
> > 
> 
> Sorry about that, uploaded again.
> 
> > Do we have a spec for structurally cloning Files? (I only see FileList here:
> >
>
http://www.whatwg.org/specs/web-apps/current-work/multipage/common-dom-interf...)
> > 
> 
> File derives from Blob, so covered by its case.
> 
> > Bumping wire format version is ok as long as you can still read the old
> > versions.
> 
> ok, the serialized-script-value test verifies that v3 formats are readable.

[jochen (gone - plz use gerrit), 2013-12-16 10:55:02]

rubberstamp lgtm

[commit-bot: I haz the power, 2013-12-16 11:54:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sigbjornf@opera.com/100023005/20001

[commit-bot: I haz the power, 2013-12-16 13:40:39]

Change committed as 163956