[Fiware-miwi] Oulu bi-weekly tomorrow at 13 again

[Jarkko Vatjus-Anttila]

FYI all,

Cyberlightning is moving to new (bigger) premises tomorrow, so we are not
able to join. Our new address starting tomorrow will be Teknologiantie 2.
See you later then!

- jarkko


On Mon, Feb 17, 2014 at 11:39 AM, Toni Alatalo <toni at playsign.net> wrote:

> Hi,
>
> welcome again tomorrow for the Oulu bi-weekly, at 13 here at
> Mäkelininkatu. Just a basic status check in the usually agile style: demos
> of things done, next plans and informing / discussion of possible problems.
>
> We can at least demo WebTundra client side physics(1) and Lasse just told
> on IRC that they've begun work on the custom components support(2). About
> others I don't know of but hopefully we'll hear latest in the meet then --
> please do post agenda items or links to code / demos at will beforehand
> too. A little info about those two items from our standpoint:
>
> (1) Physics: The old standalone three.js Car demo is ported to Tundra &
> WebTundra now, to be used with the city or any other scene. Is a separate
> project with a minimal test scene. The idea is that because physics +
> custom physics code for a lot of objects is known to be a bottleneck in
> Tundra (from avatar apps & lots of users) it can be useful for mass apps on
> the web to avoid that by using client side physics. Client side logic &
> control of the 'own entity' (avatar, car, camera, ..) is nice also to avoid
> network latency with controls. The key is that the client deals with only
> the own single object, collisions of that with nearby objects, and not
> every collision in the whole scene (like the server typically does) -- that
> makes it relatively light for each client as well. The physics is ran in a
> Web worker so doesn't block rendering. The code is on GitHub but there's no
> demo on the web as we still don't have Tundra servers up for public access
> (nor Meshmoon integration which would also solve this),
> https://github.com/playsign/WebTundraCar .
>
> This is all still custom code in that test app, to be refactored to a
> physics module for WebTundra in next steps. Then we could add support for
> reading rigidbody params from XML3D descriptions too to be able to run
> physics demos standalone (the only way currently to load a scene to
> WebTundra from a file, instead of a web socket connection to Tundra, is
> from a XML3D file -- like we use TXML with the c++ version). Also, with the
> physics module we could port the old standalone / offline version of Pong
> to use WebTundra's physics abstraction instead of Ammo.js (Bullet) directly.
>
> (2) Custom components: Lasse from LudoCraft said they've began work with
> the previously discussed plan for better custom component support. AFAIK it
> boils down to JS code becoming able to define new component types: the data
> definition and a nice way to hook custom handling of the data. The Car from
> the Physics work seems like a good candidate as a test / demo case for
> that: WebTundraCar is currently implemented with the DynamicComponent
> system so that each car entity has a Car component which is handled by the
> app code to construct the cars from the body + wheels and create the
> physics etc. They don't actually even use the Mesh component as the
> visualisation is a part of the Car component on the client side. So soon
> we'll get to see how that works with the new system.
>
> Cheers,
> ~Toni
> _______________________________________________
> Fiware-miwi mailing list
> Fiware-miwi at lists.fi-ware.eu
> https://lists.fi-ware.eu/listinfo/fiware-miwi
>



-- 
Jarkko Vatjus-Anttila
VP, Technology
Cyberlightning Ltd.

mobile. +358 405245142
email. jarkko at cyberlightning.com

Go to www.cybersli.de and enrich your presentations!

www.cyberlightning.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fiware.org/private/fiware-miwi/attachments/20140217/1bc67cd7/attachment.html>