Re: Last Call comments

10.04.2013, 15:52, "Scott González" <scott.gonzalez@gmail.com>:
> On Wed, Apr 10, 2013 at 5:57 AM, Konstantinov Sergey <twirl@yandex-team.ru> wrote:
>> 08.04.2013, 19:24, "Scott González" <scott.gonzalez@gmail.com>:
>>
>>> On Mon, Apr 8, 2013 at 11:06 AM, Konstantinov Sergey <twirl@yandex-team.ru> wrote:
>>>> Okay, look at the hammer.js pointer code: https://github.com/EightMedia/hammer.js/blob/master/src/pointerevent.js
>>>> 55 lines of code are just redundant and is used to handle the simpliest things: the number of touches and the positions of touches. And this code will be completely broken if some other script on the page uses the capturing.
>>>
>>> I think it's unfair to point to an abstraction over several event types when that's exactly what Pointer Events is. If Hammer.js was coding over the Pointer Events API and mapping Touch Events to Pointer Events, the code would be very different.
>>
>> Whether you map Touch Events to Pointer or Pointer to Touch, or event if you write code for PointerEvents alone - you have to count the pointers, to separate single-input from multi-input and to obtain the positions pointers. I could hardly imagine the application which need to work with pointers and doesn't need to do such things.
>
> In my opinion, this is a non-issue. We've already agreed to add an API for getting the full list of active pointers in v2. I'm sure I could write this in less than 55 LOC with v1. See Rick's email from yesterday [1], which mentions an easy way to create "frames".
>

Then I don't understand why to accept this v1 specifiation if it is (a) incomplete, (b) will be modified breaking back compatibility.

> I'm not even sure why this is relevant to pointer capture. Capturing to a specific element doesn't stop bubbling and it doesn't fake the x/y coordinates, so you've really lost nothing in terms of detecting gestures.

In my very first letter I stated five objections to the spec: one general objection (difficulties with implementing the basic use-cases) and four specific (capturing one of them).

This branch of conversation is about the general problems, not the capturing.

>
>> So these 55 lines of code (usually more - you usually need to dispatch the event streams because generally working with single-input and working with multi-input are completely different tasks which are solved by different modules) is an additional problem that PointerEvents add to developer because the spec doesn't bother about basic developer needs.
>
> It seems like you feel that gestures are a basic need that should be directly addressed by Pointer Events. Is that the case?
>

Just the opposite. Our Safari Mobile code use the touch* events only. We don't need gestures (i.e. specific events to detect gestures) because we just need the positions of every touch. That's one of the most useful features of Safari Mobile touch event system in my opinion.

But we failed to use the same approach in our IE10-specific code because it's very hard to use pointers without gestures. That's why I don't like the idea of discussing PointerEvents proposal alone leaving the gestures out of scope.

> [1] http://lists.w3.org/Archives/Public/public-pointer-events/2013AprJun/0048.html

-- 
Konstantinov Sergey
Yandex Maps API Development Team Lead
http://api.yandex.com/maps/

Received on Wednesday, 10 April 2013 12:41:27 UTC