Re: UA's implementation of ::selection

From: "David Hyatt" <>
Sent: Saturday, May 15, 2010 10:27 AM
To: "L. David Baron" <>
Cc: "François REMY" <>; "Boris Zbarsky" 
<bzbarsky@MIT.EDU>; <>
Subject: Re: UA's implementation of ::selection

> On May 15, 2010, at 12:24 PM, L. David Baron wrote:
>> On Saturday 2010-05-15 12:19 -0500, David Hyatt wrote:
>>> #mydiv {
>>>  selection: find-in-page(myfindstyle) selection(myselectedtextstyle);
>>> }
>>> I kind of like the idea of having only one pseudo element for each
>>> unique appearance you'd like to apply to any selection in the
>>> whole document.  Having the pseudo elements be singletons off the
>>> document root is very efficient, and then inheritance of all the
>>> selection types is handled using the same model as everything else
>>> in CSS.
>> Properties that accept value lists like that are a disaster in terms
>> of cascading.  We could never implement spellcheck or find-in-page
>> using a mechanism like that since the page would break it the second
>> it tried to style any other type of selection.
>> -David
> Yeah that's a problem.  I'd just really like us to avoid creating many 
> pseudo elements for a single selection appearance.

I have also cell-range selection in tables that styled differently from text 

I  think that we need to invent some generic mechanism that will allow
to define various pseudo-objects without need to add new selectors each
time.  At the moment we have selections, highlights of different kinds,
drag-n-drop auxiliary objects, etc.

Something like:

.parent  pseudo("text-selection") { ..... }
.parent  pseudo("cell-selection") { ..... }
.parent  pseudo("spell-error") { ..... }

Andrew Fedoniouk.


Received on Saturday, 15 May 2010 19:29:47 UTC