The following notes are taken from the TPV Developer (TPVD) meeting held on Friday, January 29th, 2016. At the time of the meeting, Second Life was experience voice issues, and so the meeting was held in text. However, the meeting was still recorded and is available here. The following notes have been taken from the chat log of the meeting.
Server Deployments – Recap
As always, check the server deployment thread for the latest updates.
On Tuesday, January 26th, the Main (SLS) channel received server maintenance package 16.01.16.310114, previously deployed to the three RC channels, comprising a simulator crash fix and feature request OBJECT_REZZER_KEY) for llGetObjectDetails(), which returns the parent_id (object or avatar) of any task in the region. The function only works for those objects rezzed in-world after the code deployment (objects in-world prior to deployment will return NULL_KEY).
On Wednesday, January 27th, the three RC channels all received a new server maintenance package,16.01.21.310177, comprising Internal simulator fixes and a fix for “LSL llSetPos in root prim of attachment behaves differently at high altitudes – viewer does not show updates”, which can cause attachments to fail to update correctly at altitudes over approximately 1500 metres.
SL Viewer Updates
HTTP / Vivox
The HTTP / Vivox RC updated to version 220.127.116.110349 on Wednesday, January 27th, which sees the addition of “Create a MailDrop event pump” to the list of resolved issues, and the removal of “[corehttp] Mac – voice does not stay connected on resume from sleep”.
The Quick Graphics RC viewer (currently version 18.104.22.1680127, dated January 20th), is getting some fixes to more accurately account for Materials, and possibly some other tweaks to how Avatar Complexity is calculated. The update should be appearing in the release channel soon.
One problem yet to be fixed within this viewer is an issue whereby avatars will still appear as solid colour jelly babies even when the Avatar Complexity slider in the viewer is set to “no limit” (which in theory should render all avatars normally). Commenting on whether or not this will be fixed, Oz noted:
The problem with those cases is that there are 2 more factors that go into making the avatar a coloured impostor, and they’re not obvious enough. We will probably change the thresholds on those, and maybe make the response to them different somehow.
When asked if those who have suitable systems will have the means to disable Avatar Complexity from running in their viewer (i.e. never see jelly babies unless the capability is expressly re-enabled) should the above issue not be resolved, Oz indicated this is a possibility, but is still to be decided.
A further Maintenance RC viewer is on the horizon, with what is described as a “nice set of fixes”. This should also be appearing in the release channel in the near future.
The Project Bento viewer continues to be worked on, and the Lab is trying to move things towards a “final” (for this phase of the work) selection of additional bones – see my separate report on the Bento project for details.
Work is all under way in updating the Oculus Rift project viewer, with an update expected in the near future. The indication is that this will be built to the 1.x SDK. Oz also indicated that the Lab is potentially looking at how many alternative HMDs they might be able to support with the viewer, including “taking a whack” at Google Cardboard support.
64-bit Viewer Builds
As noted in my last TPVD update, the Lab is starting work on 64-bit versions of their Mac and windows viewers. There is no time frame as to when the first project / TC versions might appear, but it does also mean that there will be a 64-bit viewer Havoc library made available to TPVs in due course. However, this may take a while to appear.
Currently, the Lab is using OS X 10.10.5 and Xcode 7.x for Mac builds, but are working to resolve a cmake file issue in order to build on OS X 10.11.2 or greater. This requires the Lab to re-arrange the order in which the current build tests are run to ensure the required libraries can be located as the integration tests can be successfully run.
There is still concern over Vivox voice exploits being employed by some users. One example of this is a user located on one region being able to listen in on conversations on another region without showing up on active speakers list for that region.
The problem here is that the weaknesses allowing these types of exploits tend to be within the code package supplied by Vivox, and over which LL has no direct control. They therefore require action on Vivox’s part to fix them. This is something Oz pointed out at the TPVD meeting, before going on to note that the Lab will continue to work with Vivox to get things corrected.
There are a number of changes which are being implemented throughout 2016 which will make older versions of viewers increasingly obsolete. These include further inventory improvements and the deprecation of outdated inventory messages which will impact older viewers and, most notably the switch to TLS 1.2, which is already available in the current release of the official viewer (and those TPVs adopting the CEF code base from the Lab).
As I noted (via a comment from Oz) in my last TPVD meeting report for 2015, this switch-over is being driven out of compliance requirements, rather than for any technical reason. However, what it does mean that once it occurs – mostly likely in the spring – any viewer which does not support TLS 1.2 will be unable to carry out any interactions with the SL Cashier or anything that involves money, as there will be no back-end support.
Third-Party Viewer Releases
- Catznip updated to release R10.0 for Windows only (at present) on Saturday, January 30th. This release brings Catznip back up to date with more recent code releases from the Lab, and I’ll have a review available shortly. Catznip users should note that R9 will be blocked from mid-February
- Alchemy will be releasing an update “imminently”
- Firestorm’s next release continues to look set for around early March.