13.10.2011 19:33, Evan kirjoitti:
According to 
http://groups.google.com/group/realxtend-dev/browse_thread/thread/166a75aa4a8ef0d9
assets do not need to be placed where there is HTTP access using
'localDir=' in 'HttpAssetStorage'

Ah now I know what you mean -- that localDir setting is apparently to tell the server process that it can load the files directly from disk, in the case that they are hosted on the same machine.

But also with that you still need something else, e.g. a http server, to provide them for clients.

There is currently no mechanism to transfer assets from the simulation to clients, in the client connection, but instead external connections (often to separate asset servers) are used. We know that it's different in SL originally, and transferring assets in the same knet tundra connection would be possible, but is not there now. Also Linden introduced using http for textures a while ago.

Anyhow, perhaps just using a web server works for you too? Mac OS X comes with apache that is easy to enable from preferences, Windows comes with IIS iirc and you can install apache or whatever, and is well supported in the Linux lands of course. Or?

Evan

~Toni

On Oct 13, 11:18 am, Toni Alatalo<t...@playsign.net>  wrote:
On Oct 13, 2011, at 7:07 PM, Evan wrote:

Would anyone know how to properly utilize localDir for
HttpAssetStorage?
I'm trying to run a Tundra 2.1 server with local assets at 'C:\tundra
\working\assets\' and the .txml file at 'C:\tundra\working\'
Then I'll use the client software to connect remotely via the server's
IP address.
You need to put your dir somewhere to where there is HTTP access.

Your personal webspace, or even dropbox, works for developing and testing.

Some web hosting service then for for lots of usage.

Am I going about it the wrong way by putting in --file when I run
tundra.exe for server? Is there a better way of handling import of
assets? Do I need to place the .txml file and assets elsewhere?
--file is good for the server, to load the scene. Similar to how you can load 
OAR archives in Opensimulator, if you have happened to use that earlier.

But indeed the assets need to be on the web.

The TXML doesn't need to be on the web, as the scene data that is read from it 
on the server, is what is synched from Tundra server to clients when you login 
and use the server.

I have so many questions and there seems to be a lack of current
documentation for the latest Tundra server.
Almost everything is the same as for 1 series, the change to 2 was more in the 
internals.

The new thing in 2 is that you don't need to convert asset references (e.g. 
"my.mesh") in your txml when you use it on the server, as the server can tell a 
base url like thing (default storage, the url prefix basically) during login.

I appreciate all help, even if it's a step by step.
Thank you :) .. so at least now you know that need to use some http server for 
http access -- Tundra itself doesn't have one for asset serving purposes (it 
does have some http handlers in optional modules for other purposes).

I've so far been doing the annoying asset ref conversions (with scene 
publishing scripts) that Tundra1 required, am still not fully familiar with the 
base url / relative refs / default storage thing in 2 .. but have seen the 
console messages from it when running Tundra recently, and seems that it worked 
correctly automatically for me.

Evan
~Toni

--
http://groups.google.com/group/realxtend
http://www.realxtend.org

Reply via email to