The following notes are taken from the TPV Developer meeting held on Friday, April 26th, 2019. A video of the meeting is embedded below, my thanks as always to North for recording and providing it. This was a relatively short meeting, with several periods of audio silence and text chat. The key points of discussion are provided below with time stamps to the relevant points in the video, which will open in a separate tab when clicked.
[00:00-01:17] There have been no SL viewer updates this week, leaving the pipelines as follows:
- Current Release version 184.108.40.2066190, formerly the Estate Access Management RC viewer, dated April 12, promoted April 17 NEW. – see my EAM overview for more information
- Release channel cohorts:
- Teranino Maintenance RC viewer version 220.127.116.116357, April 18.
- EEP RC viewer version 18.104.22.1686104, April 11.
- Bakes on Mesh RC viewer, version 22.214.171.1245409, March 26.
- Project viewers:
- 360 snapshot viewer, version 126.96.36.1995934, June 6, 2018 (see my hands-on update).
- Linux Spur viewer, version 188.8.131.529906, dated November 17, 2017 and promoted to release status 29 November – offered pending a Linux version of the Alex Ivy viewer code.
- Obsolete platform viewer, version 184.108.40.2060847, May 8, 2015 – provided for users on Windows XP and OS X versions below 10.7.
The EAM viewer is awaiting one simulator side update; the next RC viewer likely to be promoted to de facto release viewer will probably be the Teranino Maintenance RC viewer.
Teleport (and Region Crossing) Issues
- Major effort has been put into trying to resolve the teleport disconnect issue – hence the multiple deployments this week.
- In the process of developing and deploying the fixes, the Lab has significantly improved its ability to monitor teleports.
- This improved monitoring / stats gathering will help baseline teleports for future reference, should further issues crop up.
- It appears to suggest teleport success rates are now significantly better than prior to the updates being deployed.
- The added monitoring has no appreciable impact on performance.
Attachment Loss on Teleport
- Progress has continued with fixes for the issue of attachments coming off / becoming ghosted/ etc., as a result of a teleport / region crossing.
- It will likely be a couple of weeks before these are deployed server-side, as the Lab will be keeping an eye on the teleport / region crossing disconnect issue to ensure the fixes that have been deployed really do help break down the issues that have been experienced.
Snapshots: Flickr Cap Fail and Withdrawal of Facebook Support
Flickr Cap Fail
There is an issue with the snapshots to Flickr capability failing see BUG-226826). This appears to have perhaps been an unannounced change at the Flickr / Smugmug end of things, rather than anything the Lab has done, but investigations are still ongoing.
Update April 30th: the Flickr cap fail now appears to be fixed.
It was announced on Friday, April 19th, that support for uploading snapshots from the viewer to Facebook has now been completely withdrawn. As such, the viewer-side code is to be shortly removed, with the Lab noting:
Due to continued changes in the Facebook API, as of today the Second Life viewer will no longer be able to support Facebook Connect for sharing your in-world photos and posts. We apologise for this inconvenience and will be removing the UI from the viewer shortly. We will, of course, be happy to see your SL posts on Facebook going forward, and you can always say hello and check out what’s happening on our official page: https://www.facebook.com/secondlife.
Script Processing Issues
Issues with script processing have been raised at a number of meetings recently, and were mentioned again at this TPVD.
- Specifically, it was reported at this meeting that since the April 18th roll-back / update, some Full regions seem to have script run-time capped 12ms, so only around 60% of scripts are run, while some homesteads appear to only run around 20% of scripts..
- As it was also reported that these affected regions were running OK following the initial deployment of the newer version of the OS, this issue appears to be specifically related to the changes made on April 18th.
- The Lab is not aware of anything having been changed that might have impacted script run-time.
- The newer version of the operating system (which is specifically required as a part of preparing the simulators to a cloud-based infrastructure) is due to be further deployed to simulators Agni (the main grid) in week #18. This should provide the Lab with a broader cross-section of simulator running both the older and newer versions of the OS, which will allow a more informed comparison of metrics from the two versions to be made, which could help determine if there is a more broad-based issue with scripts on the newer OS version.
Group Notices to IM
- The Lab is considering the possibility of no longer sending group notices to e-mail when a user is off-line.
- Right now, if IM to e-mail is enabled for when a user is off-line from SL, it will allow both IMs and group notices to be forwarded to e-mail. This can result in message volumes / content being seen as spam.
- As it is invariably the secondlife.com domain that is linked to such reports, this can lead to it being regarded as a spam site, degrading the ability for secondlife.com to deliver e-mails in general, as e-mail services mark the domain as an originator of spam.
- This move is therefore intended to make e-mails from the secondlife.com domain for reliable.
- Although other forms of messaging forwarded to e-mail can potentially add to the problem of spam labelling (e.g. object IMs to e-mail), group notices are seen as by far the biggest cause.
- No decision on this has been taken either way, but disabling the ability for group notices to e-mail is easier than other options (such as requesting user go through all their groups and disable group notices).
- Were this change to be implemented, then only IMs would be sent to e-mail; group notices received when off-line would be held until the user next logs-in, as is currently the case.
- It has been suggested that increasing the limit on the number of messages that can be queued when a user is off-line might ease the blow of blocking group notices being sent to e-mail. While this idea is already under consideration, no decision has been taken either way.
[25:06-26:18] If this idea does move forward, it is hoped that changes being planned to the way SL events work will reduce the need for at least some group messages.
- No specifics were available on the updates that are being planned for events, but it has already been suggested that the updates include an API to allow events to be properly listed on viewer log-in screens (see feature request BUG-226867).
- [13:58-15:00] Following the release of a video on You Tube by a Lab employee, the question was asked about the Lab’s policy on staff using TPVs. In short:
- Lab staff must use the official viewer on Lab equipment and/or their official Lab accounts.
- Lab staff may use any viewer of their own choosing on their own hardware and when using their personal user account.
- This was actually the case with the video in question: it was filmed and released via a personal account, not an employee account.
- [31:43 (text)-35:30] Questions continue concerning LL’s support of 32-bit operating systems (it was mentioned that Microsoft will apparently stop supporting 32-bit “with the next [Windows 10?] update”). However, one of the reasons for LL’s continued support of 32-bit Windows is the number of users on less capable / 32-bt specific GPUs.
- [35:40-36:25] At least some of the EEP regressions witnessed following the April 18th roll-backs / redeployments should hopefully be corrected with the simulator updates due in week #18.
- [39:36-40:56] It’s often asked when simulator updates aren’t “more thoroughly tested” by the Lab. Simply put, such is the number of Agni (main) grid simulators coupled with the wide variety of ways people use Second Life (think of all the different in-world scripted object, or even all the purpose-built, custom windlights, for example), that replicating it all in a comprehensive test environment simply isn’t possible. Hence why the Lab use the release candidate channels: while testing is carried out (and the Lab is constantly trying to improve its test environments), the RCs provide a further means of “testing the water” before deploying updates grid-wide.
- [42:39-43:11] video playback support (MP4): work is on the roadmap for this, but the Lab has yet to get to it.