Category: TPVD / Open Source User Group

The following notes are taken from the TPV Developer meeting held on Friday, April 2nd, 2021.

These meetings are generally held every other week.  They are recorded by Pantera Północy, and her video of the meeting is embedded at the end of this report – my thanks to her for allowing me to do so – and it is used with a transcript of the chat log from the meeting and my own audio recording to produce these notes.

This as a very short meeting.

SL Viewer News

[1:13-4:55]

As noted in my CCUG summary, the Custom Key Mappings viewer, version 6.4.17.557391 was promoted to de facto viewer release status on Thursday April 1st.

The rest of the current pipeline of upcoming RC and project viewers remains as follows:

• Release channel cohorts:
  • Maintenance RC viewer – Eau de Vie, version 6.4.17.557412, dated March 25.
  • Love Me Render (LMR) 5 project viewer, version 6.4.14.556118, dated February 23.
• Project viewers:
  • Legacy Profiles viewer, version 6.4.11.550519, dated October 26,   2020.
  • Copy / Paste viewer, version 6.3.5.533365, dated December 9, 2019.
  • Project Muscadine (Animesh follow-on) project viewer, version 6.4.0.532999, dated November 22, 2019.
  • 360 Snapshot project viewer, version 6.2.4.529111, dated July 16, 2019.

General Viewer Notes

• LMR 5 is being used to address crashes relating to older hardware using Intel Graphics drivers. These have proven difficult to track down as almost nothing is reported on where in the viewer the issue causing the crash occurred. In the meantime, those on systems using older Intel HD graphics drivers are encouraged to update to more recent versions.
• Both LMR-5 and the Maintenance RC are awaiting merges with the new release viewer code.
• The Simple Cache Viewer has been updated following the forced roll-back, and should be appearing as a new RC viewer some time in the next week.
• The recent hold-up in the viewer update and release cycle means that there are now a number of Maintenance RC viewers, each with its own focus, awaiting initial release.
• A further Lover Me Render viewer (LMR-6) is also in development,  with around 8 known issues awaiting resolution before it can be considered for initial release as an RC viewer.

In  Brief

• [7:20-8:06] the re-working of the Map Tile code continues. It is believed “significant progress” has now been made, but no ETA on when the fully revised code will be in operation.

The following notes are taken from the TPV Developer meeting held on Friday, February 19th, 2021.

These meetings are generally held every other week.  They are recorded by Pantera Północy, and her video of the meeting is embedded at the end of this report – my thanks to her for allowing me to do so – and it is used with a transcript of the chat log from the meeting and my own audio recording to produce these notes.

The latter two-thirds of the meeting included a large amount of local text chat related to VRAM and texture handling. Please refer to the video for details.

End of an Era

[0.00-1:06]

As I recently blogged, Oz Linden, who initiated the TPVD meetings and who has, with the exception of when he’s been on vacation, chaired them is departing the Lab for retirement on Friday, February 26th (see: Oz Linden announces his forthcoming departure from Linden Lab). As such, this marked the last TPVD meeting he would attend, with – I believe – Vir Linden now set to carry them forward.

Given this, Oz had a few words to say to the TPV community at the start of the meeting:

 I want to say that working with the third party viewer community has been  – I mean this is how it started for me, and it has been a theme throughout, and it has been a very great pleasure working with all of you. It’s really been terrific, and thank you. Thank you for making me look good, and thank you for all you contribute to this really impactful and fun product.

SL Viewer News

[1:21-2:52]

• Current release viewer: Project Jelly viewer (Jellydoll updates), version 6.4.13.555567 and dated February 5, 2021, promoted February 17.
• Release channel cohorts (please see my notes on manually installing RC viewer versions if you wish to install any release candidate(s) yourself):
  • Love Me Render (LMR) 5 project viewer, version 6.4.13.555871, February, 18, 2021.
  • Simple Cache project viewer, version 6.4.13.555641, February 16, 2021.
  • Custom Key Mappings project viewer, version 6.4.12.553437, January 7, 2021.
• Project viewers:
  • Legacy Profiles viewer, version 6.4.11.550519, October 26.
  • Copy / Paste viewer, version 6.3.5.533365, December 9, 2019.
  • Project Muscadine (Animesh follow-on) project viewer, version 6.4.0.532999, November 22, 2019.
  • 360 Snapshot project viewer, version 6.2.4.529111, July 16, 2019.

Project Jelly Viewer

This viewer essentially improves the rendering of Jelly Doll avatars.

• Originally introduced in 2015 (and with various improvements since) as a means to allows users reduce the avatar rendering load on their systems by having any avatars around them that exceed a certain complexity value (set via a slider) render as a solid colour and minimal detail.
• There have always been a number of issues with the manner in which these avatars are rendered. For example: the colours used have been seen as intrusive so users often avoid the capability, while there have also been technical flaws such as the original code. attempting to render all of a Jelly Doll avatar’s attachments, defeating the intent of the code.
• As a result, the Project Jelly viewer improves things by both rendering avatars as simplified grey humanoid shapes, and by not making any attempt to render attachments.
• In addition it also improves to how avatar imposters are rendered and updated.
• These improvements should result in demonstrable improvements in view performance in environments where there are a large number of avatars and the capability is sensibly used.

General Viewer Notes

• Of the RC viewers, all three appear to be in good shape for promotion as the next de facto release viewer, although no decision has been taken on which will be promoted next.
• At the CCUG meeting, some users expressed a preference to see the Love Me Render 5 viewer promoted next, but there is no commitment to this being the case, as promotions are governed by stability / crash rates.
• A new Maintenance RC view is anticipated as appearing soon.

In  Brief

• [12:36-14:09] Viewer rendering:
  • There are still no firm decisions as to how the viewer rendering API will be handled in the move away from OpenGL. for the most recent information I have on this, please see my February 5th CCUG meeting notes.
  • As per my February 18th CCUG meeting summary, the current focus is on bug fixes and UI performance improvements.
  • Other proposed UI work is related to the new user experience and making the viewer easier for new users to get to grips with the viewer.
• [15:20-18:00] Chrome Embedded Framework (the media handler used by the viewer) will cease supporting Windows hardware that is pre SSE3 (2004). It is believed that few (if any) Sl users are running systems old enough to be affected by this – and if they are, they are liable to have more issues than simply losing their media playback capabilities.
• Firestorm has entered a QA cycle in preparation for what will be something of a maintenance release with a focus on closing the gap between it and the more recent Lab viewer code releases. It is hoped this will be the first in a more regular cycle of 3-monthly releases.

The following notes are taken from the TPV Developer meeting held on Friday, February 5th, 2021.

These meetings are generally held every other week.  They are recorded by Pantera Północy, and her video of the meeting is embedded at the end of this report – my thanks to her for allowing me to do so – and it is used with a transcript of the chat log from the meeting and my own audio recording to produce these notes.

SL Viewer News

[0:00-4:10]

The Project Jelly viewer updated to version 6.4.13.555567 on Friday, February 5th. This presumably brings it to parity with the current release viewer code base, and moves it closer to potentially being the next viewer to gain promotion to de facto release status, although no decision has been made on this as yet.

• Current release viewer Dawa Maintenance RC Viewer, version 6.4.12.555248, dated January 25, 2021, promoted February 1st, 2021 – NEW.
• Release channel cohorts (please see my notes on manually installing RC viewer versions if you wish to install any release candidate(s) yourself):
  • Custom Key Mappings project viewer, version 6.4.12.553437, January 7, 2021.
• Project viewers:
  • Love Me Render (LMR) 5 project viewer, version 6.4.12.553511, issued on January 7, 2021.
  • Simple Cache project viewer, version 6.4.11.551403, November 12.
  • Legacy Profiles viewer, version 6.4.11.550519, October 26.
  • Copy / Paste viewer, version 6.3.5.533365, December 9, 2019.
  • Project Muscadine (Animesh follow-on) project viewer, version 6.4.0.532999, November 22, 2019.
  • 360 Snapshot project viewer, version 6.2.4.529111, July 16, 2019.

General Viewer Notes

• With Dawa now at release status, the current RC viewers are being bought up to parity with its code base.
• As noted in my most recent CCUG summary, the Love Me Render (LMR) 5 graphics RC viewer is close to being ready for promotion from project to RC status.
• The simple cache viewer (VFS replacement viewer) is in “decent shape” for promotion to RC status once updated to the Dawa code base.
• There may be further UI work for the Legacy Profile project viewer (returns avatar profiles,etc., back into the viewer, a-la Firestorm) which may delay this viewer from progressing.
• For OS X users, a viewer is in the works that will “get Apple’s notarisations working”. This has been something of a long standing issue, and the viewer should be appearing in the near future.

Project Jelly Viewer

[9:06-11:32]

• This viewer essentially improves the rendering of Jelly Doll avatars.
• The idea behind Jelly Dolls (first introduced in 2015, with various improvements since)  was to give users the means to reduce the load of having to render extremely complex avatars on computers with low hardware specifications by:
  • Explicitly selecting a nearby avatar and set its display value to Do Not Render, reducing them to a simplified, “flat” grey avatar form.
  • And / or setting a “complexity value” within the viewer that, if exceeded by any avatar in the field of view, will render it as a Jelly Doll.
  • Both the grey and the Jelly Doll forms are simplified avatar outlines, with the latter rendered as one of a number of solid colours selected by the viewer from a colour map. The term was coined by user Whirly Fizzle after the British Jelly Babies brand of sugared jelly sweets.
• There have always been a number of issues with the manner in which these avatars are rendered. For example:
  • On the visual side, many users have avoided the viewer’s complexity setting (to the detriment of their viewer’s performance) as they do no like seeing solid, brightly coloured avatar forms in their field of view.
  • On the technical side, the code currently attempts to render all of  avatar’s attachments. As these tend to be the most costly to render, this can defeat he object of the code.
  • Also, the baseline formula for jelly doll calculations doesn’t allow for consistent results.
• As a result, the Project Jelly viewer:
  • Does not attempt to render attachments, but instead renders affected avatars as a simplified, easy-to-render humanoid shape.
  • Render all such avatars in grey, no longer using the previous colour map, in the hope this will encourage more people on lower-end systems to use the capability, as the grey avatar forms tend to be less intrusive within a scene.
• [13:38 via chat] In relation to avatar rendering / Jelly Dolls, it was asked if better global controls for rendering could be provided. In reply, Grumpity indicated that in addressing performance as a whole, such global controls might be considered in the future.

Viewer Log-in Changes

[4:41-8:05]

• Oz Linden is working on the viewer log-in process that are designed to prevent people logging-in to Second Life when their inventory is “broken” (and potentially making their situation worse).
• The updates to the log-in process means there will be additional checking on the status of an avatar inventory data on the back-end as a user logs into Second Life.
  • If errors within the data are found, the log-in process suspended, in order to prevent the errors being propagated to the viewer, and then viewer then exacerbating the situation by trying to manipulate the inventory database further on the basis of the invalid data.
  • Should this happen, the user will receive a massage that explicitly states that log-in has failed as a result of inventory issues with the request the user contacts Support. This message will also provide some specific information the user can pass to Support when they contact them.
  • Support will then use the information supplied to initiate the required corrective action to resolve the issue.
• While this may impede users with inventory problems from logging-in, the hope is that these changes will actually make the resolution of inventory-related issues easier to correct at source and thus have less of an overall impact on affected users.
• This is seen as a priority change, and the Lab hopes to be in a position to have the back-end changes tested and the viewer-side updates available by the end of the month.

In  Brief

• [4:17-4:32] The Lab is making some changes to how deployments are managed within the AWS environment. If done correctly, this work should result in no user-visible changes.
• [11:38-14:11] Post-Uplift issues:
  • There remain some issues still to be fully resolved as a result of the transition to AWS, including (but not necessarily limited to):
    • The problem with Map tile not being generated. This is being addressed.
    • The fact that the chat servers currently need to be restarted more frequently than pre-Uplift. This is still being diagnosed.
    • Teleport failures resulting in an “wrong” or “invalid” region message. This is also being diagnosed.
  • However, the Lab caution against assuming any issue that is encountered is a result of the AWS transition. The general rule remains, if you’re seeing a specific (and preferably reproducible) issue, raise a bug report.
• [14:31-15:33] in response to a question on the avatar skeleton (why 159 bones, but capped at 110 on upload?),  vir pointed out the 110 bone cap is per sub-mesh in a character, rather than on a complete character (which can have several sub-mesh components. The reason for the cap is down to older GPUs that can be used in SL being unable to handle the transform matrices.
• [19:54-20:28] Premium Plus: internal discussions have resumed on the deployment of Premium Plus, but there is nothing to share in terms of time frame, etc., at the moment.
• [21:49-24:15] The recent Marketplace issues are seen as a combination of both the continuing work to improve the Marketplace experience and the work to transition it to AWS.
  • As the MP involves multiple back-end services, there are a lot of interdependencies that can be impacted particularly as a result of the AWS transition work, and not all of these can be accurately QA’d, as the sheer volume of transaction, etc., the MP sees hourly cannot be easily or accurately replicated.
  • The current focus is on general MP stability (including its various dependencies) in order to hopefully make future maintenance and update easier / smoother.
  • The most preferable way to deal with the MP would perhaps be to take the service down entirely for a a period of time and overhaul it, and then re-release it. However, given the impact this would have, it is simply not a viable option.
• [24:42-25:45] Generally speaking, LL believe SL to be a lot more stable / robust now than previously, simply because it is running on much more recent hardware and within an infrastructure where they no longer have to worry about things like hardware that is well beyond its operational life failing, whilst any underpinning hardware / infrastructure issues are more-or-less immediately addressed by AWS. This in itself allows LL’s engineering and ops teams to be more focused on running the software side of things.

The following notes are taken from the TPV Developer meeting held on Friday, January 21st, 2021.

These meetings are generally held every other week.  They are video recorded by Pantera Północy, and her recording of the  meeting is embedded at the end of this report – my thanks to her for allowing me to do so – and it is used with a transcript of the chat log from the meeting and my own audio recording to produce these notes.

SL Viewer News

[0:00-2:15]

• Current release viewer version 6.4.11.551711, formerly Cachaça Maintenance RC viewer promoted on November 12 – No Change.
• Release channel cohorts (please see my notes on manually installing RC viewer versions if you wish to install any release candidate(s) yourself):
  • Dawa Maintenance RC Viewer, version 6.4.12.555058, January 20, 2021.
  • Project Jelly viewer (Jellydoll updates), version 6.4.12.553798, January 7, 2021.
  • Custom Key Mappings project viewer, version 6.4.12.553437, January 7, 2021.
• Project viewers:
  • Love Me Render (LMR) 5 project viewer, version 6.4.12.553511, issued on January 7, 2021.
  • Simple Cache project viewer, version 6.4.11.551403, November 12.
  • Legacy Profiles viewer, version 6.4.11.550519, October 26.
  • Copy / Paste viewer, version 6.3.5.533365, December 9, 2019.
  • Project Muscadine (Animesh follow-on) project viewer, version 6.4.0.532999, November 22, 2019.
  • 360 Snapshot project viewer, version 6.2.4.529111, July 16, 2019.

General Viewer Notes

• Viewer updates / releases had been held-up due to some Uplift related issues. This have now been rectified, and viewers should start rolling forward again.
• As noted in my 2021 week #3 CCUG summary, the Dawa Maintenance RC viewer was the only viewer updated during the week, and this will likely be the next viewer to be promoted to de facto release status.
• The Simple Cache project viewer (VFS replacement) and the Legacy Profile project viewer are both now “close” to be promoted to release candidate (RC) status. That latter had been blocked due to some required UI fixes, with Kylie Linden indicating these are in progress.
• Graphics fixes continue to be added to the LMR-5 project viewer, but this is liable to end in the next week or so, in order to allow this viewer to progress onwards through RC and eventual de facto release.

AMD GPUs “Blue View” Issue

[3:21-11:02]

Some users running systems with AMD GPU has been encountering a issue with scenes being rendered with a strong blue bias when running the SL viewer with Advanced Lighting Model (ALM) enabled and using the most recent (post October 2020, I believe, and particularly the latest Adrenalin drivers) AMD drivers.

• General reports can be found here, and a formal bug report is available: BUG229727.
• LL have acknowledged the issue, and believe it to be driver related. However, AMD driver bugs must apparently be reported using a specific reporting tool provided by AMD, and thus far LL have been unable to reproduce the bug when running said software and so are apparently having issues getting AMD to accept it.
  • The issue appears to be somewhat similar to an Nvidia bug of a few years ago, possibly resulting from a switch to using a BGRA for textures rather than the more usual RGBA format.
• There are reports that an AMD driver update issued on January 21st, 2021, but at the time of writing this summary, this has yet to be more broadly confirmed.
• In the meantime for those affected by the issue,the following workarounds are recommended:
  • Rolling back to an older driver version (e.g. Adrenalin 2020 Edition 20.11.2).
  • If that doesn’t work:
    • Disable ALM in the viewer (Preferences → Graphics → uncheck Advanced Lighting Model).
    • Log out and back into the viewer.
    • Enable ALM (Preferences → Graphics → check Advanced Lighting Model).
    • Remember to disable ALM prior to ending your session (so you only need to re-enable ALM when you next log-in.

In Brief

• [16:32-34::45] The meeting saw a significant amount of text chat concerning the technicalities of viewer CPU / CPU core usage, TPV work in trying to rebuild viewer threading, etc. As this is text-based (and may not be of relevance to many users), please refer to the video from around the 16m 32s mark through to .
• [34:47-48:00] and continuing after the brief discussion on chat lag] A further text conversation is on graphic API options and moving away from OpenGL. This is covered in my 2021 week #3 CCUG summary, but it appears that around 20% of Windows users running systems incapable of supporting Vulkan (notably those using Intel integrated graphics).
  • That said, Vulkan isn’t the sole option available to LL, and much is still in discussion internally at the Lab.
• [48:37-49:15] Some are experiencing an increase in chat lag post-Uplift. LL are currently re-starting the chat servers as an interim means of improving reliability of chat services, but it’s not clear if this is also helping reduce general performance issues.