Re: [css3-images] Changing the angles in gradients

On Wed, May 18, 2011 at 10:52 AM, Brad Kemper <brad.kemper@gmail.com> wrote:
> On May 18, 2011, at 10:31 AM, "Tab Atkins Jr." <jackalmage@gmail.com> wrote:
>> Given that I asked it, I considered it reasonably even.  It presents
>> the arguments both for CCW (matches standard polar coords, matches
>> tools) and for CW (matches other uses of angles).
>
> It didn't mention the inconsistency of this change with standard representations in leading software that has been that way for decades.

Yes it did.


> It didn't ask the people responding if they had actually used linear-gradient in a site yet (which might have informed their answers).

That's unnecessary.  In fact, I more highly value people who *haven't*
yet used gradients in any significant amount, as they aren't biased by
already training their mind in a particular way.  I have private
feedback from an author that, though they'd today vote for A because
of familiarity, they recall being confused over the way angles worked
originally, and would have voted for B or C at the time.


> You presented it as a change you were leaning towards making, and as though there would be no problems with making such a radical change now. That's biased.

There is no real problem.  All current implementations are
experimental by definition.  If pages break, they should be
maintained.  Keeping people aware that prefixed stuff may break is a
good thing.  ^_^


>> In general, it's the editor's responsibility to make the choices over
>> the specs they edit.  Especially for relatively small details like
>> this,
>
> Something that radically changes the look of every Web page currently using degrees to indicate direction in a gradient? That changes all the tools that have already been created to create cross-platform CSS3 gradients? That is hardly a small detail.

It's up to the UA to decide whether they keep their current prefixed
impl or change it, just as it's always been.  If they want to wait to
change until they drop their prefix, that's fine, as I'm trying to get
the spec into a state where they can drop prefixes as quickly as I
can.

Again, anyone using prefixed things and expecting them to be stable is
doing it wrong.  If this change happens to provide a lesson in this
principle, all the better, though that's clearly not an explicit goal.

~TJ

Received on Wednesday, 18 May 2011 18:10:23 UTC