[Opensim-dev] Any particular reason for EntityTransferModule being shared instead of per-region?

Justin Clark-Casey jjustincc at googlemail.com
Wed May 23 00:37:38 UTC 2012


In this case m_agentsInTransit.  If this is not shared between scenes then on teleport, I can put a guard against trying 
to re-establish an agent in the source region (if, for instance, some script causes an attempt to teleport back very 
quickly) before an old agent has been properly cleaned up.

One could simply look for an existing ScenePresence but then you wouldn't be able to tell if that SP was still there due 
to some crash or maybe because multiple logins are allowed and the user has some other presence there.

On 23/05/12 01:23, Diva Canto wrote:
> Hi Justin,
>
> Usually I write Shared modules if they do generic things that don't depend on the state of the individual scenes.
> Why do you want to make it NonShared? Any state in particular you are looking at?
>
> Diva / Crista
>
> On 5/21/2012 3:22 PM, Justin Clark-Casey wrote:
>> Hi Diva (primarily). Is there any strong reason for having EntityTransferModule as a shared module rather than per
>> region? I need to keep eliminating possible teleport-related race conditions and this gets simpler if scenes aren't
>> sharing the ETM module. Teleports between regions on the same simulator then better resemble teleports between regions
>> on different simulators, eliminating the need for special-case logic.
>>
>> In particular, doing this allows me to extend the in-transit checking to prevent problems such as teleporting back to
>> the same region before the old ScenePresence has been properly cleaned up.
>>
>> There are a couple of shared member variables on ETM (and HGETM) but these are either unused (m_Scenes) or I don't
>> think the share has no practical effect (e.g. m_bannedRegions, since this is only used on region cross). If these
>> shared structures are important then I could be move them somewhere else.
>>
>> In my early regression and manual tests there are no obvious issues with making this change, though I intend to test
>> more.
>>
>
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>


-- 
Justin Clark-Casey (justincc)
http://justincc.org/blog
http://twitter.com/justincc



More information about the Opensim-dev mailing list