Second Life News 2014-34

Servers

Not much going on here. The Deploy post arrived on Sunday. No rollout to the main channel this week. The RC Channels will get an update that eliminates a crash condition.

OS2014-34_002

Open Source UG – 2014-34

Viewers

The main viewer is now 3.7.14-292638. This is the Library Refresh Viewer. There is a fix in this version for gray avatars. If you have gray avatars, look in your log file for Transferred a partial file errors. They think they have it fixed, but cannot know as they have not been able to reproduce the problem in the Lab.

You can find the list of updated libraries in the viewer’s release notes

RC Viewers

As of today there are no ‘listed’ RC Viewers. However, the alternate login viewer is out there: Experimental Login Viewer. You have probably seen some pictures of it. It has the login fields at the top of the Window.

It is probably part still part of A-B testing with new users.

Project Viewers

Experience Tools – currently 3.7.12.291846 – It had not updated as of Friday, but was supposed to update soon. I can’t tell if it has or hasn’t without doing some research, which I’m not taking the time for just now.

Oculus Rift Viewer – Lab got 6± DK2’s. Oz says they will actually be able to get some work done on rebuilding the OR Viewer’s UI.

Group Chat

Is it worse? No hard numbers, but empirically it seems chat servers are locking up more often. There is an active project targeting chat, but most of that crew is on vacation and will be in week 34. If your group’s chat goes down, contact support and get them to restart the group chat server serving your group. Don’t expect change until next week.

Caching

Texture and mesh fetching – the Lab is experimenting and changing how textures and mesh items are fetched. This experiment uses a Content Delivery System (CDN). CDN’s are other servers on other Internet connections that only deliver content. They are used with lots of web sites and other web services to speed things up. It is sort-of-a-way of multi-threading the Internet.

This will take the work of finding and delivering textures and mesh off the region servers. Test regions will soon come up in the Preview Grid for testing Third Party Viewers with the new CDN’s.

Current testing is with people that have been having trouble with the present HTTP texture fetch. Their performance and reliability have improved with the new method of delivery.

For now they are not changing the viewer. If this ‘proof of concept’ experiment works then the viewer will need to change. The change would eventually lead to the removal of the process we use now. The UDP texture fetching services and code would eventually be removed from the servers.

In regard to compatibility and performance, testing has been with people in the US. They are moving to testing with some people outside the US. Then later testing with TPV, which gives a broader audience in more locations.

Early results are great.

Logging Texture Performance

The SL Viewer and TPV’s have software that logs what is happening with texture downloads. As of viewer 3.7.7 and newer, the logging is broken. That may be causing some problems. The Lab is fixing it.

By default this logging is turned off. For certain one can crash their viewer by turning it on. It requires a restart, so the viewer crashes on launch after the logging is enabled.

Hover Problem

While no one is currently working on it, there will soon be people working on fixing Avatar Height or more accurately Z-offset. So, we will be able to better adjust feet touching the floor.

The current plan is to create a new adjustment factor that is sent to the server. The server will then send that information back to all users. Very much like what used to happen with the Firestorm Z-offset adjustment around before the Server Side Avatar baking.

The current plan is the value will be secession specific, meaning when logging off the value is forgotten… which makes no sense to me and brings up lots of questions that Oz couldn’t answer.

As it is now the various things that affect Z-offset, wearables and scripts, are unpredictable. The Lab is trying to figure out a simple adjustment that would be easily understandable by most users and placed well in the UI.

The conclusion now is that there are so many things that can affect Z-offset there is no way to fix Z-offset without a high probability of breaking lots of stuff. The solution they are looking at will be a final Z-offset adjustment to the avatar-appearance-data. As it is the last adjustment and NOT part of an outfit your adjustment will work with whatever you are wearing.

Layers Thing

The Lab is still looking at how this might be done. No conclusions yet.

VS2013 & New Version of Autobuild

This is all geeky stuff about compiling your own viewer. The interesting parts from this discussion are:

The Lab now has a viewer code branch for testing 64-bit viewers. They are not going for a 64-bit build just  now. But, they are testing that changes they are making to libraries are working in a 64-bit environment. So, they are small-stepping their way there.

So, a bit of what ‘small-steps’ looks like is: as these compiler and autobuild changes move into place the Lindens will be building 32 and 64 bit versions of the viewer and all the libraries. If the 64-bit versions work, great. But, if not they will log the issue for the 64-bit version and move on to get the 32-bit version built and running.

A Significant Change in auto-build is adding clear information on the build, components used and etc., as a file in the build result.

Read more here: User:Oz Linden/Autobuild Improvements.

For those running Windows XP the change to using VS2013 (a Microsoft compiler) to build the viewer means the viewer will NOT run on XP, regardless of Service Pack installed.

Once the new autobuild is ready and put into use, XP is pretty much off the grid.

New Stuff

We aren’t going to see much. Over the new few weeks cleaning up libraries and changing over the build process is going to use up most of the teams time. Oz feels they need to stop and take time to do these things so they can move faster with fewer problems when they are working on the coming new projects.

For me this means the news cycle will be slower. I guess I could monitor the Ebola outbreak that is getting really bad.

Mirrors

Darien Caldwell tipped me off to some of the other things happening.

You probably remember Geenz Spad did the Advanced Material feature for the viewer. Well, it seems he is back adding more stuff.

See: STORM-2056Projector reflections do not respect the environment intensity parameter. This is a correction to how projected lights are reflected. The idea is to fix it and make it work better.

See: STORM-2067Glossy Projectors. This is a feature request that Geenz is working on. Geenz explains what it would do for us: This feature is important simply because projectors are becoming a bigger deal, but there’s still a couple of bugs that makes them unusable. This feature sets some ground work for glossy reflections in general as well, enabling content to look more as it should now that we have materials.

Correction: Geenz (didn’t say per Geenz – I was told he said) that with this feature implemented there would be no need for a ‘mirror’ material. It seems that this reflection thing would handle 90% of the user cases for mirrors. We may still get what some are calling Zi Mirrors, because Zi is doing the coding.

See: BUG-7023Add support for materials in basic shaders. This feature would mean that use of ALM (advanced Lighting) would not have to be enabled. Geenz describes it this way: This feature will greatly benefit the community by enabling people who previously were not able to see materials due to either lesser hardware or because they were previously unaware of how to enable ALM on some viewers. There will also be some degree of per-pixel lighting as well, however to a much lesser extent than ALM.

2 thoughts on “Second Life News 2014-34

  1. Group chat as of late is worst I’ve seen in 6 years. Thank you for the tip on contacting support about grp chat server restart. I’ll pass that along.

  2. I’ve never actually said that there would be no need for mirrors, just that there are very few use cases for realtime non-static mirrors in general. I even list some of them on my glossy reflections JIRA to address some of the comments of it. Saying otherwise generally misrepresents my opinion (which I wasn’t really asked about in the first place).

Leave a Reply

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