Second Life News Week #23

Friday was the Third-Party Developers’ UG meeting. Lots of updates on what is going on. So, check the Other News. …speaking of Other News, I am changing the order of these posts. The more interesting stuff I’ll put at the start meaning Other News. That pushes Server and Viewer reports down the page. They are still here.

The Server-Scripting UG meeting was well attended 20+. Lots of questions and answers.

Bad, Bad Panda!!

Bad, Bad Panda!!

Other News

Teleport Problems – The Lindens think they have those fixed. They have been watching the new stats they started collecting weeks ago to find the problem. It is looking good… or at least, better. If QA and RC testing had gone well, we would likely have seen the changes go grid wide this week. Maybe next week.

My experience TPing and Crossing is better. I’ve been flying and sailing with way fewer disconnects. Flying in certain RC channels I have not disconnected. From time to time in main channel regions, I do fail to connect to the destination region and get disconnected. Still, way less often.

I do still get unseated… I get left behind as my heli or boat continues on. A number of times I have recovered and caught them and continued on. Other times the controls are borked and I delete the vehicles and relog.

EEP is fighting with the fragile render pipeline. Fix one, break 2, fix two and find another one. In the Viewer section, you’ll see the massive number of fixes in the EEP Viewer.

The latest server roll gives us BUG-227100Region Windlight again broken in Simon Linden tells us that in the panic to get Disconnects fixed some fixes got dropped. They sort of passed a newer version with a Disconnect-fix version based on an older pre-EEP-fix version. Now they are trying to figure out which updates may have been dropped in the flurry and get all the new stuff in an update… basically, get back on plan.

Windlight Defaults – You may have noticed your viewer is sometimes not using the region’s default Windlight settings. This is a minor glitch in the viewers that apparently crept in as they prepare for EEP. In Firestorm, you can open Photo Tools and click Use Region Windlight to use the region’s default settings. You may have to disable and enable to get things working.

As to removing group notices from email… they haven’t started on it yet. But they will be doing considerable work on group chat, pretty soon™. The removal will be in that work.



Name changes, they are making progress. The release is not imminent but actual work is being done. So says Oz.

Attachments detaching is getting more attention. But we should be seeing less detachments as the code to reduce detachments is on an RC main grid channel. Data is being collected to see if they need to make another pass on detachments. Not knowing which RC channel, I can’t decide if it is better or not.

I seldom get off a boat or out of heli without some part of my anatomy missing. SL brings a whole new meaning to losing one’s head.

The cache system rebuild is on hold. Oz Linden plans to get someone back on that project soon™. They plan to include video ram changes along with the cache work.

The change to a newer Visual Studio is in progress. They have it building locally. But it isn’t working in the build-farm… Think of it as if I run it on my desktop machine, it works. But if I put it on automatic in the render farm it fails.

Viewer Statistics Bar – Script Performance Metrics

Script Time – If you fly or sail you may be noticing sluggish response to the controls. This is likely a problem with script time in the region. See BUG-226851Only 30 – 50% of scripts run on regions since SLS Main server roll of April 18th

Script time is the amount of server CPU time being used by scripts.

Recently we have seen a significant change in Spare Time, the unused CPU time available for scripts and other tasks. Once all the time is used up scripts wait for their time in the CPU. If a region has no time to run your heli or boat’s scripts, things get sluggish.

For discussion of what is happening with scripts see: It really is the number of idle scripts that drags down a sim.

You can check to see if it is the region or something else causing a sluggish response. Use Ctrl-Shift-1 to toggle the Viewer Stats panel open or closed. Or use the top menu Advanced (Ctrl-Alt-D) -> Performance Tools -> Statistics Bar (this is Firestorm).

The image shows the important numbers that likely indicate a script lag issue. When Spare Time and Scripts Run approach zero things are way less than ideal. In the post (link above) Animats provides the data to know when it is becoming a problem.

From the Server Scripting UG Meeting (6/4) Rider Linden: And, yes.  Idle scripts are not really idle… they can drag down sim performance even if they are not actually doing anything.  Right now llListen() is a major culprit (but I’m hoping we can clean that up in the near future)


You may have noticed we are not getting the weekly posts on the server updates. There is even a post in the Second Life Server section of the SL Forum, Deploy Plans for the week that never came.

With the TP-Crossing Disconnects peaking a week or two ago and the resulting drop in concurrent use numbers, the regular routine of write code, test, QA, RC test, and roll out on a weekly schedule seems to have been disrupted. Not surprising and likely a good thing. Think: fix it fast.

But there is also the idea of withholding information from users so as to NOT affect their bug/error/complaint reporting. Sort of a reaction to the hypochondria-like problem-solving humans are prone to. We have a tendency to think whatever changed last is the cause of our current problem, not unreasonable but… This behavior on our part has taught the Lindens they can get better reporting by NOT providing a recent change for users to focus on. I am personally involved, in a small way, so I know where some changes happened that allow the Lindens to see whether their changes reduced or eliminated the Disconnect problem. Those or maybe more accurately THAT change has not been announced. We are likely to see that change this week (23).

This lack of news leaves us in the place of having to figure out whether our problem is local, viewer, network, or system based. At least for those of us that do try to figure those things out.

The result of this is without a specific THING for the general users to point to, they point to the Lab and SL as the problem. This continues and grows to a PR problem that eventually precipitates a change in the company’s behavior. The Lab and its people being data-oriented seems to always takes things to a point where it shows in the numbers before, they react. With the Disconnects problem users were seeing it in late 2018. But it was April-May-ish before the numbers were enough for the Lindens to notice.

🎀 Post 1318

🎀 Post 1318

They don’t ignore us. But until they can see a problem, they can’t fix it or even really think about it. With Disconnects the level of anecdotal information and voices reached a point they decided they need to change glasses. Then the problem was numerically apparent.

The Lindens were poking and peeking at things months ago to see why the early reporters of Disconnects were disconnecting. Something obviously not good. But it was not a priority until the base cause was spreading across the grid and enough people were affected that millions of possibilities were eliminated to provide a focus on what the problem might actually be.

Think of fixing SL problems as like fixing the world… which it sort of is. So, what is wrong with the world?



The default viewer is version – Maintenance released Monday, May 20, 2019.  They list these fixes,

  • SL-11213Generate urls for new viewer release notes
  • BUG-226970lsl scripting touch_start() function no longer works on prims with full alpha textures
  • BUG-226982The Viewer crashes when teleporting back and forward multiple times


Release Candidates – Release Candidate viewers are being evaluated to be the next default download; all are believed to be stable.

Project Viewers – Project Viewers are experimental viewers; they may not be beta quality – use at your own risk.

The new Release Notes pages on the website are nicer. The issues fixed are easier to see. Previously the notes used the Labs internal bug report links. We cannot access those versions of the reports. The new pages use the public-external links, which we can access.

This means we can see the details in a bug report. These reports often have workarounds. Those can come in handy when we get bitten by a bug pending a fix.

Oz Linden in the Third-Party Dev UG meeting told us the Lab has received some code contributions that may allow them to put out a Linux Viewer. It will have limited functionality and certainly no voice. But the Lindens are working on incorporating it into their build system. It is a Debian flavor of some kind.

The Lab is working on a mobile viewer for iPhone. They would like to get a version for Android. They think an Android version very likely but no timeline.

Other Viewers

Black Dragon – At the Third-Party Devs’ UG meeting there was talk about the Complexity Values generated by most viewers and those generated by Black Dragon. BD can yield values 5 to 6 times greater than other viewers. Niran has used calculations thought to be more realistic. The Lab is planning to redo the complexity calculations in a project labeled ARCtan. So, they consider the current system less than ideal.

Since more places are kicking people out based on complexity some BD users are getting kicked. Fortunately, it seems rare. The servers report the complexity value to other users. But the server is told that value by the viewer. That enables our ability to set a limit on how complex and avatar we are willing to render. The server runs an average on the incoming CPI numbers. So, when the BD viewer reports a high CPI/ARC the value is discarded as out of range. So, the high value from BD generally is not a problem for the BD user. Unless a BD user is the first person in, in which case there is no average and the high CPI generated by BD is accepted as within range, the server passes that along, and the local script limiting a user’s CPI triggers and they get kicked.

For more details and explanations see: The Black Dragon high complexity (problem and solution).

Other News

This section has moved to the top of the article.


Leave a Reply

Your email address will not be published. Required fields are marked *