Post questions for Technical Town Hall with Cory Linden Dec. 2 at 4 PM PT

Please post questions about the technical side of SL on this thread for the Town Hall with Cory Linden, VP of Product Development, on Thursday, December 2, at 4:00 PM PT.

Many different posts have discussed the performance differences in sims on different age servers. It seems these statements of server affecting sim performance is valid. I have seen it on my own island sim.

Does the age of the server affect the performance of the sim? If not, how do you compensate for the older architecture on the server? And, what else can explain why the FPS on a sim changes radically when the servers are reassigned?

Thanks,
Dain

Is the possibility of a Mute User's Audio (much like the current Mute User) being considered? It seems that it's a punishment to *us* to have to mute -all- audio through preferences to escape the noise generated by one audio-griefing person playing offensive/repetitive sounds. It would be a tremendous blow against particular forms of griefing (and huzzah for that!)

Within the last few point releases, it seems as if the SL servers are attempting to send far less data than ever before. The reasons I come to this conclusion are: 1) no change in hardware or connection on my side 2) bandwidth set to default 500kbps 3) ping times always < 200ms 4) no packet loss 5) no update activity even when the near view is largely incomplete. Put another way, SL is not aggressively filling the pipe.

From prior technical explanations I expect that this could be caused by one of: a) SL has saturated its outgoing pipes b) the asset server is not capable of satisfying all requests and thus queueing or retries are happening or c) there has been a change made to prioritization of outgoing traffic to cope with a much larger grid and player base.

I know I'm not the only one to experience the "slow-rez" effect. Is LL aware of this, do they know the cause, and is there an effort underway to return it to former performance levels?

Thanks.

Are there any plans to more dynamically allocate "server" resources, other than the current static grid. That is, would Linden Lab consider a change in architecture that would allow computing and network resources to be dynamically allocated to where the AV's are at the moment, as opposed to simply allocating the same set of resources to each 256mx256m piece of land on the grid.

- Ace

What is the biggest change to SL/LSL you wish you could make but cannot, and why not?

Do you see LSL #includes, a mechanism that would allow developers to separate user-defined functions from the main script, as useful? What about a method for objects linking to scripts by key (rather than inventory) such that developers could compile scripts (in their inventory or perhaps a 'master' object) that would then be referenced by distributed objects with script keys?

A Linden-sponsored 'contrib' system for used-submitted LSL functions would allow centralized community extension of the LSL function library. Is this a direction LL would like to see LSL scripting collaboration go, and, if so, what problems would need to be addressed to make it viable?

What is the development roadmap for major SL feature changes over next quarter; where are LL's priorities? What changes are in store, particularly in HAVOK, the LSL API, and XMLRPC?

There has been much discussion about a need for more ways in which communities can understand and monitor resource usage as it relates to their 'home' sim. What challenges are posed by possible solutions and how can LL continue to extend the sim CPU resource monitoring API beyond llGetRegionFPS() and llGetTimeDilation()?

I have heard Rumors On The Internets that state it may be possible for the LSL compiler to be opened for user-contrib object code optimization, or that in some other way, further optimization may be pending. Can you substantiate these rumors and give us a picture of what may be in store down the road?

Client ATI support improvement strategies. Programmable pixel shader support. LL video driver recommendations for each major manufacturer. Would you talk about these, please?

How can LL decentralize sim hardware (or allow third-party or user-hosted sims on the grid) without also decetralizing asset management? What sort of grid infrastructure transformations can expand the SL grid beyond LL?

Has there been any progress in working with ATI to resolve issues that the Second Life client has with some ATI graphics cards? There was mention in the forums and in the support FAQs months ago regarding ATI issues, but I haven't run across any recent updates on the situation.

Some little ones...

1) Any chance of a per-sim processor cycle model for scripts/active content, in similar form to that of the prim allocation?
2) Could the geeks have a copy of the LL net topology please?
3) Is there any way for SL to split more of the render load to the GPU, dependent on the quality of the user's graphics card? (I only ask ask I've a 6800 ultra and it feels like it's being wasted...)

Many thanks.

Please could you reassure us that potentially hostile patches will be announced sufficiently long time in advance for us to roll-out corresponding patches to our own products.

Along the same thoughtline, here's my question please:

Regarding something that would be very useful for those running Photoshop or a web browser, or even in packed clubs as a balance between prettiness vs. performance... has the possibility of a performance slider in the menubar or "low-cpu hibernate" mode been considered, so that Second Life will run smoother in the background when multitasking?

Will we ever get 2-way XMLRPC, or will we ever get some other better communication mechanism for object-to-object communication?

When are you going to give us tools to detect things (objects, textures, scripts) that are causing lag in a sim so that we can tune performance in our own builds and help our neighbors tune performance in theirs?

What are the technical limitations that restrict us to using 10sqm primitives? Why not allow larger to help reduce the number of prims used overall?

I would very much like to see this addressed as well. This is a concern I have heard not only from Tony, but many other scripters... and it comes up time and time again in conversation.

What's the roadmap? Where the heck is Havok2? Will it come before the rest of the world has moved to Havok 3? Will we get the new permissions system before we die of old age?
Are you going to hire more coders now that you have a few million dollars to play with?
Are we really moving to python, and when? Will we ever be able to draw 2D GUIs/HUDs on a specific client, hidden from everyone else? Can we at least get a couple more buttons for llDialog please?

You had mentioned during the discussions about the pending permissions overhaul that you were looking at ways of making full body custom skins easier to deal with (ie making it possible for people to use a custom full body skin and still manage their own tatts). Is this in the pipeline officially and is there an ETA?

I heard a rumor that Havok 2 would become a higher priority again in December. Is this rumor unfounded?
Can we expect precompiler support in the future? (#include, #define, #if, et al)
What is the unit of mass that SL uses?
Possibilities of dynamic script memory?
Shared memory between scripts in the same sim?

How long until outbound and internal XML-RPC?

There it is again It's a popular request, this.

Is there any plans to make lights, well actually light w/o using local lighting?

Thanks,
Nathan Klein

(Im a simple man, most of the time...)

Can we get the definitive, complete description of how dwell works? Every answer I've seen is either incomplete or contradicts some other answer. And I've never seen an actual numerical formula that starts with avatar-minutes, goes to dwell points, and ends in $L.

Any more plans to improve order-of-download and order-of-render in the engine?

What sort of things stand between SL now and being able to effectively Google search SL?
Any thoughts on search-by-locality?

Any thoughts to adding new sim features? (Customizing the sky, ground, moon(s) satellites, stars, sun/light source, day/night, gravity, wind, alternate planets, etc etc etc)

Can we have a voting system built into sims that would allow sim residents to vote and change sim-wide features?

What stands between SL today and Havok 2/3 Integration?

Which is more of a user bottleneck: bandwidth, computer speed, graphics cards, or other?

1) In the sim status panel (alt-1), what is the meaning of "run tasks"? This figure seems to dominate sim CPU usage. Along the same lines, can you describe "Run Agts"?

2) You've mentioned wanting to open-source Second Life in the passed. Are there any plans to go forward with this? If so, which parts of the system will be opened?

3) In the client status panel (alt-2) what do the various items mean? What is "idle network"?

1) xml-rpc status ? when will we get outbound ? if at all.

2) linux client ? in the works or a pipe dream?

3) havok 2 eta and whats been done if any.

Linux client seconded,

Also, is there any chance on seeing user contributed LSL functions? (while the current LSL library modelled after the C stdlib, is lacking compared to languages such as Java - it would be great if developers could submit new functions for potential inclusion in future releases.)

Are any additional LSL datatypes likely to show up soon? (such as a 'long' or 'double'), and is there any chance of seeing array's or pointers within LSL anytime? Additionally, is there any hope for expanded script memory, or a permanent storage system (even if it is limited to within the sim the script is running in.)

Also, is there any hope for interface scripting, such as being able to script a completely custom GUI onto the interface. (replacing the existing one, with script hooks.), or at the very least, an improved llDialog that can act as a full dialog window with sub-widgets?

-Adam

Why does Second Life suck ass? Okay, maybe that's not a technical quiestion, but it's a technology question. Why is it that people use C# and Perl to create wonderfully useful tools, and the only things SL gets used for are malls and cars? When are we going to be able to build libraries, human intuitive wikis, and the kind of collaborative interfaces that will allow mankind to better itself?

No one asks this kind of thing. It's always ATI this, XML-RPC that... Let's talk long term goals here. When is Second Life going to change the world like the Web did, like Doom did, like ICQ did, like Everquest did, like Napster did, and like BitTorrent and RSS are doing?

When will Second Life become more than a glorified IRC chatroom? I'm bursting at the seams to use SL as a killer app, like iTunes or Apache, but so far it's more frilly crap than killer app. When is that going to change?

When I am I going to be directly link to something I've built in SL from my blog, like I can link to a Wiki entry or a forum post from my blog. When is the Second Life mantra going to stop being about immersion and start being about utility? When is it going to stop being about governments and economies, and start being about data exchange? When will Second Life mature beyond an escapist's toy and become something I can feel good about putting on my resume?

Or, the abridged question, what computer language do you recommend for someone trying to compete with Second Life? I'm leaning toward C# and .Net libraries because of the RAD capabilities. What do you think?

Inventory Blues:

1. A search function in Inventory.
2. Remove the auto-opning folder function/bug.
3. A thumbnail-view function in the Texture and Snapshot folders.

Are any of these being discussed or looked at?