Add FilePath::{As,From}UTF16Unsafe

Add FilePath::{As,From}UTF16Unsafe

R=brettw@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=207576

[darin (slow to review), 2013-06-20 19:40:36]

I put this CL together to help us eliminate
webkit/base/file_path_string_conversions.{h,cc} without incurring the overhead
of extra UTF-8 <-> UTF-16 conversions on Windows or extra typing at callsites.

[jamesr, 2013-06-20 19:50:15]

This looks good.  I've been looking at callers to the webkit/base/file_path..
functions and most of the time the data isn't really UTF-16 to start with - it's
mostly commonly derived from another encoding and then round-tripped through a
WebString.  I'm hoping to eliminate most of the callers by avoiding the
WebString intermediate since it really doesn't make sense to let Blink
manipulate paths in most cases.

[darin (slow to review), 2013-06-20 19:53:47]

Are you referring to WTF::StringImpl's ability to store Latin-1 text as a
byte array instead of as a double-byte array?  I'm not sure how to avoid
WebString in the case of WebKit::WebHistoryItem (the documentState array
contains file paths) and WebKit::WebHTTPBody.


On Thu, Jun 20, 2013 at 12:50 PM, <jamesr@chromium.org> wrote:

> This looks good.  I've been looking at callers to the
> webkit/base/file_path..
> functions and most of the time the data isn't really UTF-16 to start with
> - it's
> mostly commonly derived from another encoding and then round-tripped
> through a
> WebString.  I'm hoping to eliminate most of the callers by avoiding the
> WebString intermediate since it really doesn't make sense to let Blink
> manipulate paths in most cases.
>
>
https://codereview.chromium.**org/17509002/<https://codereview.chromium.org/1...
>

[jamesr1, 2013-06-20 20:21:45]

No, I mean things like the database paths which are computed in ways that
can only produce only ASCII, or paths that go into Blink but probably
never should like the path to plugins.  It sounds like some cases are going
to have to stick around, though.

[darin (slow to review), 2013-06-20 20:48:36]

OK... yeah, makes sense.


On Thu, Jun 20, 2013 at 1:21 PM, James Robinson <jamesr@google.com> wrote:

> No, I mean things like the database paths which are computed in ways that
> can only produce only ASCII, or paths that go into Blink but probably never
> should like the path to plugins.  It sounds like some cases are going to
> have to stick around, though.
>

[brettw, 2013-06-20 21:24:10]

lgtm

[commit-bot: I haz the power, 2013-06-20 21:26:55]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/darin@chromium.org/17509002/1

[darin (slow to review), 2013-06-20 21:30:00]

Committed patchset #1 manually as r207576 (presubmit successful).