Hittest finds the truncated text instead of the floating input, when the input is clicked.

[zalan]

<style>
  div {
    text-overflow: ellipsis;
    overflow: hidden;
    white-space: nowrap;
  }

  input {
    float: right;
  }
</style>

<div><input type='text'>long long long long long text</div>

1, input is clicked.
2, text gets the hit test, even though it is truncated right before the floating input box.

attached test case

[zalan]

Created attachment 200808 [details]
reduced test case

[zalan]

1. We need to use the m_truncation at InlineTextBox::nodeAtPoint() to make sure truncated text is ignored while hittesting.
2, We need to teach the parent RootInlineBox to be smarter at hittesting (InlineFlowBox::nodeAtPoint()) to figure out that the truncated part.

[zalan]

[12:57am] dhyatt: if i were writing this feature from scratch i would probably have let the child elements overflow
[12:57am] dhyatt: but shrink the root line
[12:57am] dhyatt: might be some reason why i didn't do that though
[12:58am] dhyatt: easiest fix at this point is to just hack it then prob
[12:58am] dhyatt: painting must just be hacking it
[12:58am] dhyatt: so hit testing can too

[zalan]

<rdar://problem/13008605>

[zalan]

Created attachment 201179 [details]
Patch

[zalan]

will be moving layout test to /fast/css

[zalan]

Created attachment 201456 [details]
Patch

[Dave Hyatt]

Comment on attachment 201456 [details]
Patch

Make a vertical-rl test too.

[zalan]

Created attachment 202237 [details]
Patch

[zalan]

(In reply to comment #8)
> (From update of attachment 201456 [details])
> Make a vertical-rl test too.
added (best effort). Vertical text truncate bug is tracked here: bug #116413

[zalan]

Created attachment 205277 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 205277 [details]
Patch

Clearing flags on attachment: 205277

Committed r151894: <http://trac.webkit.org/changeset/151894>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.