From the Third Party Viewer meeting we learn that the viewer install code has been changed to make use of RC and Project installs easier for the Lab and TPV Dev’s. Earlier I mentioned Name RC Viewer as having the program code changes that allow viewers to be more easily named. Being able to better control channel names cleans up some install problems as well as data tracking.
I am taking what I hear Oz Linden saying about not wanting to get into the details of this change to mean it is complex with Linden preferences driving some of the changes. As long as things work better, none of the detail matters to me.
The Maintenance RC Viewer had a really bad crash rate. It has been withdrawn and users pushed to other viewer versions. If you installed it and disabled automatic updates, you need to replace it.
The Interesting Viewer has a horrible crash rate. Fixes are being made. The ‘Interesting’ change is a great success, things render faster. We may see an update before the no change windows arrives. The next update should be much better.
This viewer is a case of a viewer that looked very good in Linden QA testing. But, it did not stand up well in RC testing. The verity of machines, video cards, and builds encountered in RC testing revealed problems not seen in QA.
The Fitted Mesh Project Viewer is not getting much use. That means there are too few users to generate reliable crash statistics. Something less than 2,000 users. What numbers they have, while not providing reliable test data, seem to indicate the viewer is neither too bad or too good.
In the latest release of Fitted some bone handles were removed and some added because they were not working. Some skin weighting was improved to work with the new bones. The changes should not affect existing skins or animations.
When the Fitted Mesh viewers are released, old viewers that do not know about Fitted Mesh and/or the new skeleton will render Fitted Mesh clothes oddly. This means people will be needing to update viewers to remain compatible with SL. Old mesh and liquid mesh clothes made previously will still work. But, new clothes that use the new bones will render oddly on old viewers.
The Lab is getting good feedback on Fitted Mesh and the new skeleton bones. They are changing things based on the feedback and suggestions. Oz says, “Things are in flux. Changes can still be made.”
More feedback would be helpful. If you know people that are doing rigging and making mesh clothes now would be the time to get them testing it and providing feedback (JIRA FITMESH). This is the window in which users can make a difference and change how fitted mesh works. Once the window closes, getting changes made will be much more difficult. Oz says it will likely be years before the Lab is willing to make avatar changes again. So, contact those you know and point this opportunity out. Do it now.
The Lab considers the new skeleton subject to change at any time. Thus it is not yet ready for general use. We will know the Lab has decided the skeleton is in the final form when we see the viewer move into an RC channel. At that point I suspect the window for making changes will be pretty well closed.
Trivia: Why did it take the Lab so long to decide on Fitted Mesh? Oz says the reason is the people that were needed to study the problem, make a decision, and move the project forward were involved in Server Side Appearance/Baking. SSA was the priority.
Once the right people were involved, it didn’t take long to make a decision.
Oculus Rift Viewer is in internal testing. Oz says has seen the team making various builds of the viewer. But, as yet it has not yet made it to an RC or Project channel.
You probably know that Second Life™ voice is provided by a third party named Vivox. Vivox is removing some seldom used parts of the service (something about presence data when they were testing voice from outside SL into SL) the seldom used features and data create performance problems. Viviox is dropping those features. This in turn means that the Lab is removing those parts from the SL system.
In week 50, if not now, those changes should be running on all of the ADITI grid (preview grid). TPV Dev’s that have viewers relying on those aspects of Viviox’s voice server will be testing viewers.
The changes will roll out on the main grid some time after the first of the year.
As far as I know there is no longer any feature in the SL Viewer that relies on the features being changed in Vivox.
Brook and Don Linden provided some information on AIS.
You probably know this as the Inventory APIv3. Load testing is complete on the Lab’s side. They are preparing for a pile-on-test. They expect the real change to be with inventory operations that fail in APIv2. Those types of failures will be caught by APIv3 viewers and then run to completion.
Firestorm has an AISv3 Beta Viewer in the hands of their beta testers. For the pile-on test the Lindens hope to get 40± testers. TPV Dev’s are working out when they can have 40 or so testers available.
Brook was thinking if all went well, testing was accomplished in week 50, and results were good there would then be a chance this could be ready before Christmas. Considering the beginning of the next no change window is week 51, I don’t see how server side code would make it through the RC channel before the New Year. But, maybe I miss understand what Brook means by ready.
TPV Dev’s doubt they can get 40 beta testers for a pile-on this month. So, this is most likely going to be something for next year. The details are being worked out by TPV Dev Teams with Oz Linden.
For technical information on the API see: Inventory API v3.
No Change Windows
If you don’t already know, no change windows are periods of time where the Lab runs SL using minimum staff. If something goes wrong, people have to be called back from holiday to get it fixed. To avoid that, the Lindens avoid rolling out new viewer or server software. No changes are made that might mess things up.
In addition they request TPV Dev’s not release changes or updates that might create problems.
The coming no change window is Dec. 16 to Jan 1. Since January 1 is a Wednesday there won’t be any server roll outs that week. My thinking is the Lindens need to back in the office and making decisions and it will a day or two after the 2nd before we see activity. So, effectively the practical no change window from a users point ends the 5th.
Software libraries are collections of programming code that handle often-needed tasks. For instance file open tasks or establishing an Internet connection. Some are written by the Lab, but many are written by third parties. Some come from Microsoft for Windows systems, some from Apple for Mac’s, and various sources for Linux.
All of the various libraries are called from the programs used to build the Second Life systems. Those programs that build programs are referred to as compilers. They combine all the code written by the software engineers and the authors of libraries to make the final ‘compilation’ we know as the Viewer or Simulator software. Thus the name ‘compiler’ for the software that makes software.
Compilers can read scripts that tell them how to compile a program to run on Windows, Apple, or Linux. The scripts can be complex. The less orderly the arrangement and handling of libraries the more complex the scripts become. So…
This ‘library clean up’ project started because Monty Linden needed to make an update in the cURL library. Things were out of date and confusing. So, he has ended up updating the libraries and how the Lab will handle libraries in the future. This also changes how the viewer-build-system works and is organized.
This is part of the path preparation that will allow easier preparation of 64-bit viewers. But, the immediate purposes of the changes are for the purpose of getting HTTP Pipelining working. Pipelining is a way to reduce the work a computer or server needs to do to connect and transfer data.
Said another way, when Monty started to make the changes needed to get HTTP pipelining working he found a mess. Lab’s libraries had been built as needed. The overall style of managing libraries sounds as if it were pretty much non-existent. Whether or not there was a prescribed style the result had been lots of inconsistencies and confusion.
Monty is cleaning things up and asking for TPV Dev help in deciding the best ways to make some of the changes. The result will be a cleaner library system that is for everyone easier to understand, use, and update/maintain. That will, as a byproduct, also make moving to 64-bit viewers easier.
There are new server API’s coming. Some of the API work is to get the Linden Scripting Language (LSL) editor updating its syntax files from the servers. It is still in testing/QA but should see an RC channel in January.
Syntax files control the highlighting and tool tips the viewer’s script editor uses. This change will mean the viewer’s script editor will always be up to date with the state of server side LSL. Previously the SL severs would get scripting language updates and then the viewers would have to be updated. The viewer’s could lag several weeks behind. Now the viewers can read the updates from the servers.
When building things people have wanted to be able to set default permissions on them. The servers are getting an API that will allow viewers to communicate those preferences to the server. So, we will be able to set default permissions.
No new news. Baker Linden is still working on changes to Group Bans. We have no ETA.
TPV Dev Meetings
No meetings for the rest of the year. Next meeting will being week 1 of 2014.