The following notes were taken from my chat transcript + the video recording by Pantera (embedded at the end of this summary) of the Third-Party Developer meeting (TPVD) held on Friday, June 6th, 2025. My thanks to Pantera as always for providing it.
Meeting Purpose
- The TPV Developer meeting provides an opportunity for discussion about the development of, and features for, the Second Life viewer, and for Linden Lab viewer developers and third-party viewer (TPV) / open-source code contributors to discuss general viewer development. This meeting is held once a month on a Friday, at 13:00 SLT at the Hippotropolis Theatre.
- Dates and times are recorded in the SL Public Calendar, and they are generally conducted in text chat.
- The notes herein are a summary of topics discussed and are not intended to be a full transcript of the meeting.
Official Viewers
- Official Viewer: 2025.04 – 7.1.14.15192634334, issued May 25, promoted May 28 – NEW.
- Chat Mentions (Early Support): Type @ then pick a name. To follow: audible alerts and highlight colour pickers. This does not support generic mentions such as @everyone or @here.
- My Outfits subfolders: now supports the use of subfolders.
- Build Floater improvements: increase to scale boundaries; Physics Material Type now updates when selecting linked objects; Repeats per Meter value no longer incorrect for non-uniform sized objects.
- Hover height: the minimum/maximum is now +/- 3 meters.
- Snapshot floater: L$ balances can be hidden independently of the rest of the UI.
- Preference Search bar: general usability and readability improvements.
- Refer to the release notes for full updates and fixes.
- Second Life Project glTF Mesh Import, version 7.1.14.15361077240 June 2 – NEW.
- This is an early Alpha release with some of the rough edges and already resolved many bugs and crashes, although more are to be found, together with general feedback from the community. Please read the release notes if you intend to test this viewer.
- Second Life Project Lua Editor Alpha (Aditi only), version 7.1.12.14888088240, May 13 – No Change.
glTF Mesh Uploader
- The GLTF Mesh Uploader will be the next viewer release, although it doesn’t as yet have a new-format version number.
- It is now available on the Alternate Viewers Page (and the version current at the time of writing is linked-to above).
- The initial release is regarded as a “pretty rough cut”, with fixes already streaming into the code branch such that there is likely to be an update within the week.
- Those trying the initial release should be aware that rigged mesh import in particular is in an “iffy state” for some of the more complex avatars.
- As the next planned release, it will be merged-up to the Develop branch shortly – so if anyone would like to get some PRs in now would be a really good time to do that.
Test Items
Linden Lab is seeking content content creators who would be willing to contribute items for testing. Specifically, the Lab is seeking:
- Rigged meshes only – unrigged content is not required at this time.
- Everything from the most basic avatar mesh to complex bento, fitted, etc. meshes, clothing and Animesh is welcome.
- However, avatars with onion layers should not be submitted. If possible, merge such avatars into a singular mesh with rigging in order to help simplify the Lab’s debugging work.
- Content should be supplied in both glTF and COLLADA.DAE formats, to allow for A/B testing.
- Content files should be sent to gltf@lindenlab.com.
glTF Uploader “Phase 2”
- A “reunification” of the various asset import flows, focused on under-the-hood work with the import pipeline.
- It used to be that assets like sounds, animations, images, etc., were uploaded through a single API when uploaded, but since the arrival of mesh, new asset upload types have diverged from the “common” flow to use their own.
- The idea now is to provide a single API to get things from a creator’s computer and in-world to simplify the upload process.
- According the CCUG meeting, this work will include an improved preview capability, providing more representative of what creators can expect to see under a set of EEP parameters, etc.
- The ability to preview items in-world on Agni (the main grid) prior to upload will not be provided. However, there is still Aditi (the beta grid) for this.
- It is also hoped that the import flow can be made easier to understand for existing and new content creators (e.g. providing a better preview; having things laid out in a way that makes potential problems more obvious, etc.).
In Brief
- Switching away from OpenGL: this has not been mentioned for a while, and currently, it looks like LL is going to use an existing API abstraction, so as to allow Apple Metal to be targeted as well. The API mentioned was “something based off of NVIDIA’s Slang API” – with a note that hopes of getting “something potato friendly” is low.
- However, there has been no firm decision, and alternatives were suggested in the meeting, so things are still up in the air.
- Signal Linden requested feedback on the idea of deprecating and removing Navmesh characters / llCreateCharacter (i.e. Pathfinding).
I don’t state the idea lightly, and I understand its low adoption is partially due to a complex and buggy implementation. However, I pulled some data and it appears we have 1,900 navmesh characters across the entire grid. Half of those are ours (Think LL experiences.) There are 6 user-owned regions with more than 10 characters… This is very low usage. I haven’t seen any major merchants release notable content with the system due to its unreliability: but please correct me if I’m wrong.
– Signal Linden
-
- Those at the meeting didn’t object to the idea.
- An alternate approach (“more nuanced” as Signal Linden referred to it, once raised) suggested was to leave server-side the functionality in place and just drop the viewer side, until such time as new ways to handle Pathfinding could be developed.
- Signal repeated that there are benefits (Pathfinding accounts for a lot of simulator-side code), but that not decision has been made as yet, he is seeking feedback.
- The subject of Havok in the viewer was raised (used for both Pathfinding and mesh decomposition), with suggestions to replace Havok with the open source HACD library or VHACD (or similar newer implementation), were this to be done.
- A suggestion was made for TPV developers to bring a HACD / VHACD to the viewer as a code contribution.
- The above lead to a wider discussion on textures (slightly sidetracked by a complaint over the cost of 2K texture uploads for non-Premium + members), and related issues of texture loads on GPUs with limited VRAM, texture crushing, etc., which continued through the latter part of the meeting.
Next Meeting
- Friday, July 4th, 2025 at the Hippotropolis Theatre – but check the SL Public Calendar for possible changes.
† The header images included in these summaries are not intended to represent anything discussed at the meetings; they are simply here to avoid a repeated image of a gathering of people every week. They are taken from my list of region visits, with a link to the post for those interested.
Inara Pey: Living in a Modemworld 