Re: [css3-images] Summary of recent gradient issues

On Jun 16, 2011, at 1:31 PM, Tab Atkins Jr. wrote:

> On Tue, Jun 14, 2011 at 4:42 PM, Simon Fraser <smfr@me.com> wrote:
>> This confusion of starting point vs. angle is specious.
> 
> I know you recently did a Twitter poll to gauge thoughts on the
> matter.  I conducted my own and got substantially different results.
> I believe this is because of the way are two questions were worded.
> 
> Yours was "Straw poll: if you see linear-gradient(left, black, white),
> do you think "starts at the left", or "ends at the left"?".  Mine was
> "Poll: Given that 0deg points up, do you expect
> linear-grad(bottom,black,white) and linear-grad(0deg,black,white) to
> be the same or opposite?".  Hopefully you'd agree that my question
> wasn't leading, but it was intended to explicitly hit the "Are these
> consistent when considered together?" angle.  I also posted the same
> question with 'top' instead of bottom, so people could answer that one
> if they felt stronger about it.  (I can tell who responded to what by
> seeing which they replied to.)  I also posted a followup tweet
> specifying that I was specifically looking for feedback from people
> who hadn't yet used CSS gradients much or at all (I know that your
> poll received at least a few answers from people who were very
> familiar with current CSS gradients).
> 
> I received a total of 14 responses from Twitter, and 2 privately:
> 
> * 2 people thought that 'bottom' and '0deg' were the same (current WD)
> * 12 people thought that 'top' and '0deg' were the same (current ED)
> * 2 people thought that 'top' and '0deg' were the same, but that they
> both pointed down (opposite of current ED)
> 
> (I wonder what the result would have been if I said "0deg goes from
> bottom to top".  I suspect that would have shifted it at least
> somewhat more towards current WD.)
> 
> So, from comparing your survey results and mine, we can learn a few things:
> 
> 1. Thought of by themselves, the keywords make people think of
> starting position.
> 
> 2. In concert with angles, the keywords make people thing of ending position.
> 
> 3. By themselves or in concert with keywords, angles make people
> either think of the gradient direction or the ending position (it's
> difficult to tell which one most people are thinking of, as they
> produce the same conclusions).
> 
> So, I think the main conclusion we can draw from this is that the
> keywords are confusing when contrasted with angles, while angles are
> pretty universally understood in any context (at least once people
> learn which direction the angles are facing, and we've already
> established that the current text matches most people's
> preconceptions).
> 
> This leans me more strongly towards either dropping keywords entirely
> and using one of the angle-only solutions (1, 2, or 3 in the OP), or
> switching to keywords that are unambiguously directions
> (upwards/rightwards/uprightwards?)

I wondered about new keywords too:
  rightwards, leftwards, upwards, downwards.
or the suggested "left-to-right", or "left to right" notations.

I think you'll have  a hard time persuading people that they
need to write out "left to right" when just "left" could work though.

I agree that my straw poll wasn't realistically comparing the angle syntax
with the keyword syntax, but that wasn't the intent. The intent was to 
gauge whether "linear-gradient(left, black, white)" should ever mean
anything other than black being on the left. The response was overwhelmingly "no".

Simon

Received on Thursday, 16 June 2011 23:39:50 UTC