[Ur] Ur/Web & RDP

[David Barbour]

On Sun, Feb 26, 2012 at 9:38 AM, Adam Chlipala <adamc at impredicative.com>wrote:

> I don't think you need to be very clever to do what you want with RPCs and
> message-passing.  Don't worry; there is no per-event polling involved.
>  (I'm not suggesting you use periodic tasks at all for this sort of thing.)
>  If you look at the demos of message-passing (these are the last few in the
> main demo site), either (a) it should quickly become very clear how it can
> work for you or (b) you can point out documentation problems that prevent
> it from becoming clear. :)


I could probably update a representation of an RDP graph within each
transaction, which would more closely match your use of message passing.
But I expect it wouldn't scale well to many clients; there'd be a lot of
transaction conflicts, and redacted communications due to new updates in a
round. I think this simply isn't a very close match for the Ur/Web
framework as it exists.


>
>  Something like http://jsfiddle.net/jashkenas/**CGSd5/<http://jsfiddle.net/jashkenas/CGSd5/>,
>> with a tunable number of elements in a page, might be useful to demonstrate
>> animation of your reactive system. The latency question could be answered
>> by animating, say, the motion of a mouse pointer through the server across
>> frames.
>>
>
> In the design of Ur/Web, I haven't focused on "animation" at all, but
> rather on conventional page content trees that evolve in response to
> events.  I don't think the current Ur/Web supports "animating mouse pointer
> motion" in any way that is at all pleasant without using the JavaScript FFI.


I just meant moving an image or div around corresponding to the motion of a
mouse pointer by reactively specifying its position. The latency, and
variability in latency, is the interesting question with regards to the
mouse. Visual motion information would be superior to text in revealing
failures.

Ur seems still worth pursuing, but for now I've decided to focus on a
simple Haskell implementation of RDP, and maybe work through your CPDT and
Pierce's SF to develop a formalization in Coq.

Regards,

Dave
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.impredicative.com/pipermail/ur/attachments/20120226/2d2bacea/attachment.html>