Server Deployments – Recap
As always, please refer to the server deployment thread for the latest information.
- The was no Main (SLS) channel deployment or re-start on Tuesday, April 11th.
- On Wednesday, April 12th, the three RC channels all received a new server maintenance package which includes:
- Several internal fixes and two new internal logging modes
- Another adjustment to fix issues with off-line IM and Group Notice delivery reliability
- Fixes an issue where large numbers of objects could be returned after a rolling restart.
Upcoming Server Updates
Note: DRMSIM is the Aditi (beta grid) channel reference.
DRTSIM-323: New Simulator Build
This update was delayed in its deployment to a release candidate channel on Agni (the Main grid), with Mazidox Linden indicating it will now be moving in week #16 (commencing Monday, April 17th). Region holders wishing to test their applications / services etc., on the new simulator build should contact Concierge about having their region moved to the appropriate RC channel(s) when known.
DRTSIM-332: Updated Avatar Capacity / Access for Regions
See either Improved Region Capacity and Access from the Lab or Lab announces improved region capacity and access in Second Life in this blog for details. This update is currently on a micro-channel (McRib) on Agni, with Mazidox noting the following:
After careful consideration, we’re making the new limits a default, rather than requiring intervention from Estate Managers (of course it is still possible to set limits lower using the region console or the current Maintenance viewer at http://wiki.secondlife.com/wiki/Linden_Lab_Official:Alternate_Viewers ). As a result there’s a new version on Aditi and we’ll have an updated version for McRib soon.
DRTSIM-343: Allow Public Access Region / Parcel Settings Changes
This is the updates to region / parcel access that will mean that if a region is explicitly set to Allow Public Access, parcel holders on the region will no longer be able to override the setting at the parcel level (see my update here). The update had been deployed to the three RC channels, where it was awaiting a viewer-side update featuring revised / improved land controls. However, Mazidox notes the code:
Was removed from RC after residents raised concerns about losing their previous state of “Allow Public Access”. To alleviate concerns of accidental change by landlord wiping out existing access settings on all parcels, we’re saving parcel setting until the region restarts. This means that while it won’t be saved permanently, it will be restored if the landlord removes their Ban Lines override. DRTSIM-343 will likely be back this week in testing, but may not have a corresponding viewer change yet.
DRTSIM-347: Fix for Incorrect Object Returns
This fix is designed to prevent a bug wherein objects might be returned from a region when it is restarted. It is about to undergo testing on Aditi prior to onwards deployment to Agni.
Content Creation User Group
The following notes are taken from the Content Creation User Group meeting, held on Thursday April 13th, 2017 at 1:00pm SLT at the Content Creation User Group wiki page. The audio extracts are from my own recording, and a video of the entire meeting, livestreamed by Medhue Simoni, is embedded at the end of this update.. The meeting is chaired by Vir Linden, and agenda notes, etc, are available on the
HTTP Asset Fetching
Vir re-iterated that the HTTP asset viewer, which utilises HTTP and the Content Delivery Network(s) leveraged by the Lab to deliver all Second Life asset types to users (rather than just mesh and texture assets and avatar baking data) as is currently the case with release versions of viewers), is now at Release Candidate status. The viewer has gone through two rapid-fire updates, with the latest version (at the time of writing) being 188.8.131.525368. I have an overview of the viewer, which includes an explanation of CDN use, for those who may not be familiar with the way things work.
Applying Baked Textures to Mesh Avatar
[Video: 2:04-18:35] The Lab is digging further into this idea, which if implemented would in theory allow at least some texture layers to be baked onto avatar meshes in a manner akin to that used with system layers and the default avatar. This could greatly reduce the complexity of mesh avatar bodies by removing the need for them to be “onion skinned” with multiple layers, although there are some issues which would need to be addressed.
One is that the baking service cannot currently handle materials (normal and specular maps; only texture – diffuse – maps), and the Lab has no idea as to how widespread either might be among clothing makers specifically producing applier systems for mesh bodies (which can support the use of materials). As such, they do not have a clear idea as to whether support for materials through the baking service would be required; something that could make any project related to this idea far more complicated to develop.
This led to an extensive discussion on what would be required from a system which could bake directly onto mesh layers, how widely materials might be used, and exactly how such a system would be used. The broad consensus of the discussion was that in terms of baking down layers, most creators like likely prefer to see things like skins and skin variations, tattoos, lingerie, freckles and eyebrows baked down to a single layer, none of which tend to require significant materials support.
The discussion also touched upon the opportunity to present better alpha masking / alpha blending than chopping up mesh bodies into sections for masking / having alpha blending on different layers get into a fight, etc. Overall, it was felt that presenting some means to bake down some layers and reduce the complexity inherent in mesh avatars would be preferable to waiting on the Lab to be able to undertake a more widespread overhaul of the baking service to provide “full” materials support through it. Vir is going to take these points raised back to the Lab for further consideration.
Animating mesh objects is another project under consideration. This would be especially useful for things like non-player characters (NPCs) and the like. There are several ways this might be approached, but as Vir explained in reply to a question, none of them would involve animating the Collada file. He also pointed out that animated objects is still only at the discussion stage, so whether any project – were it to be taken on – would include the ability for animated objects / NPCs to attach static objects (e.g. a hat or similar), hasn’t been addressed.
Animating Objects (/ NPCs): This is also still under consideration, but yet to be adopted as a project. However, were it to go ahead, it would likely not involve animated the Collada files, as suggested during the meeting, but leverage SL’s existing animation capabilities. It’s also too early in the process to say whether or not animated objects would support static objects being attached to them (e.g. an NPC being able to wear different hats).
Supplemental animations: The idea is to allow “supplemental” animations to run alongside the animation states keyed by llSetAnimationOverride(), effectively allowing them to play together, rather than conflicting with one another as is the case at the moment. Suggested some time again, it is still be considered, but no work has been carried out as yet.
Rendering Cost Calculations: he Lab is running a background project to look at the cost of rendering a wide range of Second Life features across a range of different client systems running the viewer. The tests for this work have now been defined and are about to be put into use. Once sufficient data has been gathered, the Lab will use it to determine what might need to be done to improve the accuracy of the avatar rendering calculations. Vir further defined the project’s status:
Re-using Linden Water Maps: The question was asked about re-using Linden Water data (Linden water being a combination of animated diffuse and normal maps) so that it might be re-used on other surfaces. The obvious use here would be to enable mirrors. Quite aside for the feared performance hit this could cause, it seems likely that the water system probably stands as its own unique implementation which would not easily lend itself to other uses.
Bouncing Bewb Improvements: In creating a set of animation for a client, Medhue Simoni noticed an issue with breast attachments and rigged mesh breasts. Essentially, as breast attachments go to the chest, they don’t necessarily follow the movement of the breasts when physics are used (which utilise the volume bone). His suggested solution would be to allow breast (/ nipple) attachments to be attached to the volume bone instead, allowing them to naturally follow breast movement,
Aditi Inventory Server Update
The Lab has upgraded the hardware for one of the Aditi inventory servers. Almost all users logging-in to Aditi should be using it. There is a request that if those who do file a bug report f they notices anything strange related to inventory: lag, failure to load, textures looking incorrect etc.