Use SubSelector as relation on left-most selector

Use SubSelector as relation on left-most selector.

Selectors used to end with a Descendant relation in the left-most simple
selector. That didn't make any sense and it caused an accidental reset of
the adjacent flag in selectorList traversal during feature set updating.

R=chrishtr@chromium.org
BUG=435923
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=185966

[chrishtr, 2014-11-24 18:32:27]

Are you referring to the issue mentioned in this FIXME?

https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

[rune, 2014-11-24 21:52:23]

On 2014/11/24 18:32:27, chrishtr wrote:
> Are you referring to the issue mentioned in this FIXME?
> 
>
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

Yes. You're thinking about making sure we set SubSelector on the last component
instead? Sounds tempting, might be a lot of code to change, I don't know.

[chrishtr, 2014-11-24 21:53:21]

On 2014/11/24 at 21:52:23, rune wrote:
> On 2014/11/24 18:32:27, chrishtr wrote:
> > Are you referring to the issue mentioned in this FIXME?
> > 
> >
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...
> 
> Yes. You're thinking about making sure we set SubSelector on the last
component instead? Sounds tempting, might be a lot of code to change, I don't
know.

Or a new None more like. It'd be nice to fix this design bug at its root. Give
it a try?

[rune, 2014-11-25 10:52:08]

On 2014/11/24 21:53:21, chrishtr wrote:
> On 2014/11/24 at 21:52:23, rune wrote:
> > On 2014/11/24 18:32:27, chrishtr wrote:
> > > Are you referring to the issue mentioned in this FIXME?
> > > 
> > >
>
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...
> > 
> > Yes. You're thinking about making sure we set SubSelector on the last
> component instead? Sounds tempting, might be a lot of code to change, I don't
> know.
> 
> Or a new None more like. It'd be nice to fix this design bug at its root. Give
> it a try?

I don't think it makes sense to add a separate enum for the end case. It's given
by a null tagHistory anyway. SubSelector is not a combinator. I think a better
naming would be to rename relation() to combinator() and rename SubSelector to
None. That's quite intrusive and should not be done as part of a bugfix, I
think.

I've tried to fix this by using SubSelector at the end now, modified the
description, and removed my initial fix. Let's see if it passes the tests.

[chrishtr, 2014-11-25 17:44:57]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2014-11-25 17:44:57]

lgtm

[commit-bot: I haz the power, 2014-11-25 17:45:23]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/751023002/40001

[commit-bot: I haz the power, 2014-11-25 19:35:49]

Committed patchset #3 (id:40001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=185966