Re: Gradient syntax proposal

On Aug 15, 2009, at 7:56 AM, Tab Atkins Jr. wrote:

>> I'd much rather drop bg-position (in case you couldn't tell by now)  
>> and just
>> have one word to say what direction to fill the box (sometimes  
>> implied by
>> the name of a side or corner, sometimes the other way around), and  
>> no more
>> than one "how far along that direction" per color. That would take  
>> care of
>> 99.999% of the author's needs and be far, far simpler. I really  
>> can't see
>> the compelling need to be able to specify a starting position  
>> different from
>> first color's position, or an ending position different from the  
>> ending
>> color's position. THAT makes it much harder to read and more  
>> confusing.
>
> I think you may be slightly confused about how the <bg-position>
> construction works.

Nope.

> It doesn't specify "a starting position different
> from the first color's position".  It just sets where the 0% point is
> (and the 100% point, for the other one).

I know what it does.

> Assuming you're setting your
> starting point at 0%, or omitting the length so that it defaults to
> that, then the first color will start exactly where the <bg-position>
> specifies.

Yes. So it is redundant. You are creating a second area that  
percentages refer to instead of the box. And when people start copying  
other people's code to learn this, they will see that bg-position is  
set for the two outside points, and that for other points in between  
you use percentages between those points. You'll have two ways to do  
the same thing, depending on how a person learns it. And that will be  
the predominant way that people use beginning and ending positional  
numbers in combination with color-stop percentages. Just muddying the  
waters. And very, very few will actually use it for your use case, of  
intentionally setting a sub-area of the box to do the >0 – <100%  
transitions within.

> As well, the simple use-cases you want *are already covered*.

And so are the more complex use-cases, well enough, with the simpler  
form. Good enough that anyone looking at the results would not be able  
to tell the difference. They are all that is needed, and it is not  
worth adding the confusion, complexity, and redundancy for those  
extremely few cases where it wouldn't be EXACTLY the same.

Received on Sunday, 16 August 2009 17:29:29 UTC