[Opensim-dev] REST and OpenSim

[Dr Scofield]

Frisby, Adam wrote:
>
> I've been evaluating replacing BaseHttpServer completely with a new 
> class based on the C# Webserver (HttpServer.dll) I committed a few 
> revisions back.
>
is that the one from  http://www.codeplex.com/webserver?

    cheers,
    dr scofield/dirk
>
>  
>
> The reasoning being that BaseHttpServer requires Administrative rights 
> under Windows, and is the only component of OpenSim that does so -- 
> this severely restricts our ability to secure windows systems running 
> OpenSim. The secondary reason is that we cannot programmatically use 
> SSL certificates via BaseHttpServer -- something the revised one 
> supports. It also gives us a good deal more control over the output 
> than we currently have.
>
>  
>
> Regards,
>
>  
>
> Adam
>
>  
>
> *From:* opensim-dev-bounces at lists.berlios.de 
> [mailto:opensim-dev-bounces at lists.berlios.de] *On Behalf Of *Alan M Webb
> *Sent:* Friday, 16 May 2008 11:44 AM
> *To:* opensim-dev at lists.berlios.de
> *Subject:* [Opensim-dev] REST and OpenSim
>
>  
>
>
> This is not intended to imply anything about the long-term strategy 
> for BaseHttpServer, rather it offers a tactical solution that 
> addresses a specific need that those of us trying to implement REST 
> interfaces have today.
>
> As REST becomes more prevalent, there is a growing need to accommodate 
> it, and its special requirements, in BaseHttpServer. I'd like to 
> suggest that we do so by incorporating the notion of a User-Agent 
> sensitive handler. The handler is registered using the name of the 
> User-Agent it intends to support. This handler receives both request 
> and response objects and returns nothing (to BaseHttpServer), i.e. it 
> is wholly responsible for header management and everything else, as is 
> the nature of REST. Along the way we might want to implement some 
> generic REST support to make implementing handlers easier, but it is 
> not required. A generic handler might take care of "hiding" the 
> request and response packets in the way that Stefan has suggested.
>
> I have a version of BaseHttpServer which incorporates these ideas 
> without affecting any traffic not associated with the specific user 
> agents being targeted. Note that the idea here is not to provide agent 
> handlers for common agents such as Firefox, but rather to insulate 
> traffic from a functionally specific agent from conflicts with other 
> generic flows; REST is just one beneficiary.
>
> I have attached a patch that illustrates the changes I'm talking about.
>
>
>
> Thoughts? You can incorporate the change without affecting the way 
> that OpenSim works today, and only if you add and agent for a specific 
> client, does anything change. The client dictionary uses its keys to 
> detect duplicate entries, but the actual matching can be arbitrarily 
> complex. In fact, what if we allowed an agent to match based upon an 
> arbitrary examination of the request headers?
>
> Best regards
> Alan
> -------------------
> T.J. Watson Research Center, Hawthorne, NY
> 1-914-784-7286
> alan_webb at us.ibm.com
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>   


-- 
dr dirk husemann ---- virtual worlds research ---- ibm zurich research lab
SL: dr scofield ---- drscofield at xyzzyxyzzy.net ---- http://xyzzyxyzzy.net/
RL: hud at zurich.ibm.com - +41 44 724 8573 - http://www.zurich.ibm.com/~hud/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20080519/d696daba/attachment-0001.html>