Tuesday 18th December saw the Main Release channel get the maint-server release as deployed to Magnum in week 50, as Maestro Linden indicated might be the case during the Server Beta meeting on the 13th December. There have been some reports of teleport issues following the deployment, but these are thought to be a network-side issue, rather than a problem with the server code itself.
All three RC channels should be receiving the same package on Wednesday 19th December, 2012 which is also the Magnum deployment from week 50, with a single additional bug fix.
Release notes (for Magnum, but applicable to all three RC channels)
This is the last set of server-side deployments for 2012.
A low-key Simulator User Group meeting on 18th December, the last for 2012
SL Viewer Updates
The release version of the official SL viewer rolled to the 3.4.3 code on Tuesday 18th December, with the release of 3.4.3.268262. Among a range of updates, this release include the major GPU table updates seen in recent beta viewer releases, the additional graphics settings and the tiling fix for high-resolution snapshot (MAINT-638). This release also sees the return of JIRA numbers to the release notes, matching the recent beta viewer release notes update.
The new “intermediary” graphics settings intended to better represent the default SL capabilities of different GPU classes
The beta viewer should be moving to the 3.4.4 code with the next release, although it is not clear whether this will be before or after the holiday period.
Interest List Update
Andrew Linden has been tweaking and optimising his Interest list code. The first phase of the new interest list code is aimed at reducing the amount of information being sent to the viewer by the server. Speaking at the Simulator User Group meeting on the 18th December, Andrew commented, “Last week I managed to find an optimisation opportunity in my Interest List project, which shaves a few hundred usec off of some cases where there are a lot of Full Object Updates. So this would translate to better server FPS when in large groups of avatars or when in the presence of stuff that is generating lots of full updates.”
As I reported last time, object updates take three forms, Full, Terse and Delete. Full updates are generated when an object is seen for the first time, or which is undergoing a constant change in appearance / position. Terse updates are sent only when an object changes its appearance / position. Under the current Interest list system, both types of update are sent to the viewer, regardless as to where the viewer actually needs to process them. A primary part of Andrew’s work has been reworking the code so that updates – particularly terse updates – are only sent to the viewer when the object generating them is in the camera’s field-of view. This new optimisation work should further help reduce the number of updates being sent to the viewer regardless as to whether or not it has to process them.
Andrew Linden continues to forge ahead with his initial work on Interest Lists, which forms a part of the Shining Project. As reported last time around, the issue of people’s HUDs appearing on other people’s screen has been fixed, and the code is currently with LL’s QA for testing.
Given the progress made, Andrew re-capped on the project / gave further insight into this initial phase of the work Server Beta User Group meeting on Thursday November 29th.
Object Updates
The first phase of the new interest list code is aimed at reducing the amount of information being sent to the viewer by the server. This done through the code only sending required updates to the viewer for objects which are within the camera’s line-of-sight. Essentially, updates on objects take three forms regardless of which Interest List code is being used:
A “full” update – required when an object is “seen” for the first time or which is constantly updating position / appearance
A “terse” update – required only when an object is changing appearance / position relative to the viewer’s in-world view
A “please delete” update when the object has been removed from the viewer’s in-world view (e.g. it has been deleted or taken back to inventory).
The new updates are aimed specifically at the number of “terse” updates being sent to the viewer. Under the current Interest List code, these updates are continuously sent out by the server to all viewers in range of an object in motion or undergoing change, regardless as to whether what is being updated (in terms of movement or appearance) is actually visible in the window of the viewer. With the new Interest List code, updates are only sent to your viewer based upon what is actually in your world view.
This means, for example, that if you can see a bouncing ball on your screen and then turn your avatar or camera so it is no longer visible to you, under the old system, data packets relative to the ball’s motion continue to be sent to your viewer, even though they are no longer required. With the new system, the updates cease shortly after the ball moves off your screen, and only resume as the ball moves back on to your screen once more (no actual content is broken by this change, it is simply a change in the amount of updates being sent from server to viewer).
The net result of this is to reduce the amount of data the server is sending to the viewer, thus helping improve performance. As the new code applies to both objects and avatars, this can amount to a substantial improvement, as Andrew commented during the meeting, “I think we’re currently seeing a 30% improvement (in the time spent in “Agents” in the stats) for the case of about 30 avatars running around in a region with 12k prims”.
However, there is a side issue with these changes, which Andrew and the devs are currently looking into. If you turn away from a moving object for a length of time such that the terse updates from the server are no longer sent, then suddenly pan the camera / turn so the object is once again in view, there might be a brief delay (one second, currently) before the object correctly updates. This is because the viewer is rendering the object based on its “old” data received from the server before getting the latest updates from the simulator. The time delay can potentially be reduced, but doing so can negatively impact the overall performance gains made. Because of this, Andrew is holding it at one second in order to ascertain how much of an impact the delay actually has as the code is tested.
Camera Follow
Currently, everything you see in-world is almost exclusively based on its position relative to your avatar, regardless as to where you move the camera. This is why, as you cam further away from your avatar, objects may appear with less and less detail, or may not render at all (particularly smaller objects) – their respective level of detail is being calculated based of their distance from your avatar, not from their distance from your camera.
With the new Interest List code, what you see in-world is now based upon the position of your camera. The benefit of this being that as you cam around, objects should render at the correct level of detail relative to your camera (so no more sculpts which appear to be stuck at a lower LOD despite your camera hovering a few metres away from them, for example). The difference between the two approaches can be seen in the image below.
Interest List in action: In both images, I’m standing 100m from a .5-cube with a 0.001 black cube on top of it. When I cam out to the cubes using the existing Interest List code (top image), the black cube fails to render, due to the level of detail sent to the viewer is based on my AVATAR’S position, despite my camera only being a metre or so away from the small cube. However, under the NEW Interest List code (bottom image), the small black cube is rendered, because the level of detail being sent to my viewer is now based on my CAMERA’S position (click to enlarge)
After indications from LL that there may not be a Main channel deployment on Tuesday 27th November, restart commenced as the deployment made to the RC channels last week went ahead as per the usual schedule.
Wednesday 28th November should see the three main RC channels updated as follows:
BlueSteel and LeTigre: should receive a maint-server project. There are a few new flags for the LSL function llGetObjectDetails(), but the most important changes are some fixes for physics and mesh-based crash modes – see the server release notes
Magnum should receive the say package, with additional stability improvement changes – see the Magnum server release notes.
As usual there is a forum thread for the week’s server deployments.
Viewer News
Release Viewer
The 3.4.2 viewer code finally reached the release (production) version of the LL viewer with the release of 3.4.2.267137 on Monday 26th November, which I briefly reviewed here.
Beta Viewer
The beta viewer, now cleared of the crash issue bottleneck, moved rapidly through the 3.4.2 code base prior to Thanksgiving in the US, as previously reported in the news updates, and then reached 3.4.3 with the surprise release of 3.4.3.267135 during Thanksgiving week, after it had been indicated there would be no viewer releases during the week due to decreased support staff availability during the long weekend period. As reported last week, this release includes the first phase of Monty Linden’s HTTP texture fetch project, which should see people experiencing significantly faster texture rezzing when in-world.
CHUI Viewer
The CHUI – the Communications Hub User Interface – project viewer is due to go through another couple of iterations before moving towards a development / beta viewer code merge. There has already been one update since the project viewer, which is aimed at improving the capabilities and reliability of in-world text and Voice conversations, first appeared.
CHUI: potentially a couple more iterations to come
While he has not followed the project first-hand, Oz Linden believes CHUI to be nearing a “feature complete” status. The advice is that if you haven’t tried it out and wish to give feedback, now is the time to do so.
Nalates Urriah provides an update on some of the ongoing work around the mesh deformer. In the meantime, speaking at the Open Development User Group meeting on Monday the 26th November, Oz linden responded to a question from White Rabbit as to what garments are still required for testing by saying, “That’s a great question. I’m setting up a meeting with the people responsible for avatars to try to get a proper acceptance test defined for both that and STORM-1800.” STORM-1800 relates to the vertex weights of the default avatar character mesh.
While Oz didn’t specify a date for the meeting, those with a direct interest in either supplying mesh clothing for testing or in the JIRA should be hearing from him in the near future on the meeting details.
Wednesday November 14th saw the same code deployed to all three RC regions in preparation for the US Thanksgiving week code freeze (see below). This primarily brought all three RCs to the same code level, release-wise and fixed a bug discovered in week 45.
There will be no rolling restarts in week 47 (week commencing Monday 19th November) due to the code freeze.
SL Viewer Update
The beta viewer rolled to the last of the 3.4.2 releases on Thursday November 15th, with the release of 3.4.2266995. Providing the crash statistics remain good (they were very positive for the first 24 hours), it will be going to the production (release) viewer QA in week 47 and should result in the release of a new version of the viewer shortly after the Thanksgiving weekend.
This beta contains a lot of updates and improvements, as well as a wide range of fixes for issues encountered with earlier releases up to and including the previous beta release, 3.4.2.266708, issued on November 8th. For a comprehensive list of changes, please refer to the release notes.
At the same time, the development viewer rolled to 3.4.3.267061, marking a further update of the development viewer to the 3.4.3 code, which should be appearing in the next release of the beta viewer. This include the new viewer-side code for the HTTP texture fetch project developed by Monty Linden as a part of the Shining Project improvements.
The code for faster texture fetching / rezzing has moved from a project viewer into the viewer-development stream and is present in the 3.4.3267061 Ddevelopment viewer release and should appear in the beta viewer after the US Thanksgiving weekend
As the holiday period is approaching, viewer releases will be slowing down, but the aim remains to try to clear the backlog of waiting merges and updates by the New Year with a view to resuming their more usual cycle of releasing a development / beta update around every three weeks. Once things are back on track, LL will be looking more closely once again at the question of disabling tcmalloc completely within the viewer.
Code Change Freezes
The official dates for code change freezes during the upcoming holiday periods currently stand at:
Week 47 – Monday 19th November through to Sunday 25th November
Week 51 – Monday 17th December through Sunday 23rd December
Week 52 – Monday 24th December through Sunday 30th December.
No status has yet been given for the New Year week (Monday 31st December through Sunday 6th January 2013.
Avatar Baking
Bake fail: work on new service progressing
On Friday 16th November, Nyx Linden provided a brief update on the Avatar Baking project, which again forms a part of the Shining project improvements to Second Life. In short:
The viewer code is starting to look stable. However, merging the new code into the existing viewer code is liable to be somewhat more “painful” (Nyx’s term) than had been hoped
Work is progressing on the server-side elements (the composite baking server), with the code reaching a point where avatar texture can be generated server-side
Currently, the plan is to continue working on both sides of the equation, with the aim of ensuring the new code is successfully merged into the viewer development branch, and then offering it in a “very alpha” form to TPVs so that they can start merging it into their code and assist with testing. As this happens, one or two regions on the beta (Aditi) grid will be set-up to allow testing on the new service.
There is still no time frame for the appearance of the viewer code in the development branch or any test regions on Aditi, but in Nyx’s view, both are liable to be on the horizon “pretty soon”. Overall, the plan still remains to have at least a two month period from when the code is made available for testing purposes through to the official implementation of the new service.
Interest List Demonstration and Weirdness
The focus of this project is to optimise the data being sent to the viewer, information already cached on the viewer and the manner in which that data is used in order to ensure it is used more efficiently so that things rez both faster and in a more orderly manner than is currently the case.
This work is being undertaken by Andrew Linden in a number of stages, the first being to clean-up the code related to information sent to the viewer from the simulator relating to objects in the camera’s viewing range such that only objects actually in the camera frame are updated (if updates are required) and that objects outside of the current camera frame are not updated, thus reducing the amount of data both the server and the viewer need to process, which should lead to performance improvements.
It is possible to visualise the update process using a viewer debug setting (Develop > Show Info > Show Updates to objects or Ctrl+Alt+Shift+U) to show object updates being received by the viewer. This shows updates in three colours: red, which indicates the viewer is receiving a “full” update on the object (generally because it is being “seen” / is within draw distance) for the first time; blue, which indicates the viewer already has data on the object and is receiving “terse” updates relating to changes in the object’s appearance / position relative to the viewer’s camera position; and green, which indicates the object has been deleted / remove from the camera view, and updates are about to cease.
On the 15th November, Andrew used this debug setting, together with a set of scripted “bouncing” cubes to demonstrate his improvements to this update process. Observers were asked to focus their camera on the cubes, which were initially static and had no coloured data stream associated with them. The cubes were then set bouncing, which generated a stream of blue “terse” updates, indicating the motion of the cubes in the viewer was being updated. However, when observers angled their camera to view the space above the cubes (so the cubes were not directly in their world view), the update stream ceased – indicating the viewer was no longer receiving update data for the cubes.
This may seem a small change, but it does dramatically decrease the amount of information the viewer has to process relating to in-world objects, and should result in performance improvements within the viewer. In the future, further work will be undertaken to enhance the interest list code even further – such as prioritising the order in which objects in the world view are actually rezzed, so that items closest to the camera view are rezzed first, etc.
HUD Issues
Following the demonstration, however, some people started noticing an odd issue: they could right-click on the centre of their screen and reveal a prim attachment belonging to someone else’s HUD. Chieron Tenk was the first to raise the issue, although Ana Stubbs also quickly reported the same problem.
Chieron Tenk posted an image of the problem: a prim appears in the centre of his viewer which, on inspection, appears to be linked to a HUD worn by Rex Cronin
After initial confusion, investigation revealed it was possible for anyone to find they had random prims from other people’s HUDs appearing on their screen, simply by attaching a HUD or a prim to their own screen. Concerns were further raised when it appeared that events might be able to be triggered if the prim was touchable.
I find I have a prim belonging to Ana Stubb’s Mystitool appearing on my screen
The issue appears to be tied to changes made to the interest list code on the test region, and is obviously going to be the subject of further investigation on the Lab’s part prior to the interest list project being carried forward.
Deployments for the week are progressing as planned.
Main Channel
The main channel received the code which had been running on the Magnum RC channel as well as some updates.
This now means that the server-side Group Services code to improve the loading and editing of very large groups (10K+ members) is active right across the main grid – see the section on Group Services below for further information.
This roll-out restored functionality within the Estate Tools which allows region physics to be put in a condition of limited functionality, which is sometimes useful in dealing with issues and problems within a region. The capability was disabled around the time of the mesh roll-out, and has now been restored with this release. This caused some minor inconvenience on some regions (at least one), where the option has been enabled at some point, with the result that following their restart, objects within the region(s) were not functioning correctly. However, this was corrected without major incident.
Wednesday 13th November will see all three Release Candidate channel receive the same update package, including the BUG-166 update, which means that linksets with bounding boxes larger than 64m (in any dimension) are prevented from being rezzed if doing so will cause the object to collide with an avatar excluding the object owner.
There will be no server roll-outs in week 47 (week commencing Monday 19th November) due to the forthcoming Thanksgiving weekend in the United States. There will be deployments between Thanksgiving and Christmas, but it currently looks as if these may be limited to a couple of weeks during that period according to Simon Linden (details on the likely number pending), and there will as usual be no releases over the Christmas / New Year period.
Interest List and Object Caching
Andrew Linden reports that the first phase of this work is drawing to a conclusion, and he is planning on having a possible demonstration of the capability on the beta (Aditi) grid on Thursday 15th November, most likely during the Bet Server User Group meeting.
The focus of this project is to optimise the data being sent to the viewer, information already cached on the viewer and the manner in which that data is used in order to ensure it is used more efficiently so that things rez both faster and in a more orderly manner than is currently the case.
Interest lists and object rezzing: initial srver code updates almost ready
Andrew reports that general performance on object rezzing should be improved, although the overall sorting element of the code (ensuring objects closer to an avatar’s camera position rez sooner than those further away) isn’t currently as rigorous as it could be. However, the server and viewer do now interact better, so less information is sent to the viewer relating to in-world items which are not visible within the current camera view for the viewer.
Commenting on demonstrating the capability when speaking at the Simulator User Group meeting on Tuesday 13th November, Andrew acknowledged that it may be difficult to achieve on Aditi, which is a relatively static environment (improvements will hopefully be more noticeable on regions where there is a lot of movement and activity); however, anyone interested in this work may want to try attending the Beta Server Group meeting on the 15th November, in case a demo is provided.
Currently, this represents the fist pass in Interest list improvements, and one which is liable to be heading to an RC channel in the near future – it does not require any specific viewer updates to work -, and Andrew expects to be building on this work in the future.
Group Services
As mentioned above, Baker Linden’s Group Services HTTP code is now available across the main grid. As there was some confusion evidenced on Plurk yesterday, here’s a quick re-cap on what this means:
The new code allows for improved loading of membership lists of very large groups, together with improved reliability in editing such groups (i.e. assigning roles, removing people, etc.), by the group moderators
The code requires a viewer update. At the time of writing, this is available with the official Second Life beta viewer (3.4.2.266708+), and the code will be filtering into the majority of popular TPVs as they update (and currently appears to be available in Zen (3.4.2.2+) and Niran’s Viewer (2.0.3.2262+) and Cool VL (1.26.4.38), all of which successfully loaded large group lists for me)
Until such time as the viewer-side code has been incorporated into TPVs, the “old” method of loading group lists into the viewer will still be available. However, viewers using the “old” method (a protocol referred to as UDP) will have group loading capped at 10K members. This means:
That for groups with 10K or fewer members, there will be no change regardless as to whether the viewer is using HTTP or UDP
But for groups large than 10K, viewers running the UDP code will be unable to load the group until such time as they have been updated to the new code
The code will not lead to any improvements in group chat reliability, and is not aimed at improving group chat.
Materials Processing and Avatar Baking
No news on either of these, beyond what has been previously reported in these pages. Materials processing has a test region on Aditi, but there is no timeline on when a project viewer is to be made available. For an overview of the initial capabilities for material processing, please see my project update here, and remember that the capabilities will be applicable to prims and mesh, but not directly to avatars or system layer clothing.
Avatar Baking is progressing, but without any significant update at this time, please refer to my last detailed update on this project for information.
Mesh Importer Fix
JIRA SH-3055 records a problem with the official viewer’s mesh uploader which has been affecting people over the course of the year. The fix for this, released as a project viewer (3.4.2.266471, available for Windows, Linux and Mac OS) on November 5th, is still available for those experiencing uploader issues, although it is in the pipeline to be merged with the beta viewer now that crash issues seem to have been resolved. Bear in mind that – as Runitai states in his JIRA comment, the viewer is a pre-beat project version, and may include other bugs and problems. While reports on the JIRA seem to point to it being relatively stable, caution should still be taken if attempting to use it as a primary viewer.
The deployments schedule for this week (Tuesday 6th and Wednesday 7th November) went ahead as planned, namely:
Tuesday 6th: the Main Channel received get the code currently running on BlueSteel and LeTigre – release notes
Wedneday 7th:
Magnum received get fixes and updates to the code currently running there (including the Group Services code) – release notes
LeTigre and BlueSteel should get the next bug fix server in the pipeline, which includes the code currently on Magnum, and more – release notes (BS) and release notes (LT)
The main channel deployment now means that all regions are running on the same version of Havok with the exception of Magnum regions, which should be getting the update in week 46 (see below).
LeTigre and BlueSteel both have an additional “feature”: Linksets which have bounding boxes larger than 64m (in any dimension) are prevented from being rezzed if rezzing would cause the object to collide with an avatar excluding the object owner (BUG-166).
In addition, both LeTigre and BlueSteel include the following oft-requested bug fixes:
Script Time in the Statistics Bar now correctly shows 0ms when scripts are disabled in the sim (BUG-311)
Script error messages now include information about the object’s root prim, when certain operations fail due to the object’s pathfinding setting (PATHBUG-198).
A crash bug was also found in the Magnum code, and this has received attention, with the fix due to go out next week.
Week 46 Deployments
Things are gradually slowing down in preparation for the Thanksgiving code release freeze which will see a suspension of code deployments during the Thanksgiving week later in November. As it stands, the following roll-outs are planned for week 46 (week commencing Monday 12th November):
Main channel: should receive the code currently running on Magnum (including Baker Linden’s Group Services code – see later in this article)
Magnum: should receive the code currently running on BlueSteel and LeTigre, which will mean the entire main grid is now running the same version of Havok
BlueSteel, LeTigre and Magnum should also get the same additional updates (details yet to be specified).
Beta Viewer Update – Steaming Ahead with Project Code Merges
As indicated in Part 1 of this report, the crash issues impacting the beta viewer code have been resolved, and LL have been engaged in merging-up code into the beta and paving the way for the first of the 3.4.2 beta releases. These were always intended to have the code from some of the major SL projects which impacted the viewer, including Baker Linden’s Group Services code and Monty Linden’s HTTP texture fetch code.
The first 3.4.2 beta viewer was release on Thursday 8th November (3.4.2.266708), which includes a range of updates from the Lab as well as a number of contributed updates and improvements (see the release notes), although precisely which of the LL project elements are in the release isn’t obvious from the release notes themselves – the removal of JIRA numbers from the release note entries makes identifying updates, features and fixes that much harder, even though the JIRA items themselves are still open for public viewing.
One element that is clearly in the latest beta viewer release is the code for the steam link-up, as evidenced by the arrival of the new splash screen which I first reported on back in August 2012 – complete with a promotional piece for the Lab’s Pattern’s game.
Inara Pey: Living in a Modemworld 