Re: [cssom-view] Which APIs should be 'double' vs. 'long'

On Wed, 02 Dec 2015 09:42:13 +0100, Trav Stone <travtrilogi@gmail.com>  
wrote:

> Hi,
>
> Jumping in this thread; it would definitely be helpful to have some API
> that returns the exact values that the browser is operating upon. See the
> following bugs I posted:
>
> https://code.google.com/p/chromium/issues/detail?id=559245
> https://bugzilla.mozilla.org/show_bug.cgi?id=1226695
>
> In my use-case, what's of interest is knowing (via JS) when the browser  
> is
> generating an ellipsis. Most of the time this works with integers, but
> there are cases where it fails. Unfortunately in the application I'm
> working on there's sufficient data being displayed to almost guarantee  
> that
> the problem will surface on any given page.
>
> Also worth noting: in my use-case I'm working with text in a TD element,  
> so
> getClientBoundingRect isn't an option unless I want to add extra markup.
> Given that our application is so data-intensive (with corresponding  
> markup
> complexity), extra markup is a fairly onerous burden.
>
> I understand the issues with changing the int values of the existing
> API(s), so I am strongly in support of making a new API to resolve the
> issue.
>
> Many thanks!

OK, thanks for your feedback.

A proposal for getting accurate scroll* information:

[[

partial interface Element {
   [NewObject] DOMRect getScrollRect();
}

1. Let document be the element’s node document.
2. Let rect be a new static DOMRect with x/y/width/height set to 0.
3. If document is not the active document, return rect and terminate these  
steps.
4. Let window be the value of document’s defaultView attribute.
5. If window is null, return rect and terminate these steps.
6. If the element is the root element, follow these substeps:
    1. Set rect's x to the x-coordinate, relative to the initial containing  
block origin, of the left of the viewport, or zero if there is no viewport.
    2. Set rect's y to the y-coordinate, relative to the initial containing  
block origin, of the top of the viewport, or zero if there is no viewport.
    3. Let viewport width be the width of the viewport excluding the width  
of the scroll bar, if any, or zero if there is no viewport.
    4. Set rect's width to max(viewport scrolling area width, viewport  
width).
    5. Let viewport height be the height of the viewport excluding the  
height of the scroll bar, if any, or zero if there is no viewport.
    6. Set rect's height to max(viewport scrolling area height, viewport  
height).
    7. Return rect and terminate these steps.
7. If the element does not have any associated CSS layout box, return rect  
and terminate these steps.
8. Set rect's x to the x-coordinate of the scrolling area at the alignment  
point with the left of the padding edge of the element.
9. Set rect's y to the y-coordinate of the scrolling area at the alignment  
point with the top of the padding edge of the element.
10. Set rect's width to the width of the element's scrolling area.
11. Set rect's height to the height of the element's scrolling area.
12. Return rect.

]]

This is the same information as  
scrollLeft/scrollTop/scrollWidth/scrollHeight, but without the quirks mode  
nonsense (the body never maps to the viewport here), and x/y/width/height  
are unrestricted double. Thoughts?


In your case, I think it should be possible to use getBoundingClientRect()  
on the cell; it should give the same result as the above API for a table  
cell.

To measure the width of the *text* in the cell, without extra markup, you  
should be able to use  
https://drafts.csswg.org/cssom-view/#extensions-to-the-range-interface

Something like  
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3789

<!DOCTYPE html>
<table><tr><td width=100 id=cell>asdf</table>
<script>
var range = new Range();
range.setStartBefore(cell.firstChild);
range.setEndAfter(cell.lastChild);
w(range.getBoundingClientRect().width)
</script>

-- 
Simon Pieters
Opera Software

Received on Friday, 18 December 2015 10:48:37 UTC