Friday (2/26) saw the Third Party Developers user group meeting. It lasted about 40 minutes. We have some interesting news and even a few specific details. You can watch the video of the meeting here, thanks to Pantera Północy.
There isn’t much change in the Viewer pipeline. Work is proceeding but some of the problems have taken time to resolve. The RC viewers are revealing problems, which is the whole point of having RC versions.
Oz Linden believes they gave solved the biggest problems in Second Life HTTP update Viewer version 184.108.40.2061302. This version has what developers refer to as Aura’s Inventory Updates. Aura’s inventory update will let them start removing old inventory API stuff. Then they can disable/remove the supporting server side code. This will kill non-Aura-compatible viewers.
Oz says the HTTP feature will be a difficult merge. This means when these change are merged into the main viewer’s code they will have issue to resolve. If you have ever synchronized mail boxes, folders, or computers, you have run into problems where the sync tool didn’t know which item to keep, duplicate, or delete. Sort of the something in merging, but a bit more complex.
Oz says one the HTTP viewer is out they will start working on HTTP access path for gestures, animations, sounds, then UDP paths for those items will be removed.
There is bug in sound engine that causes it to miss playing a sound first time it is asked for. This problem is unlikely to be fixed by changing to HTTP deleivery of the sound file.
Going to HTTP protocol will make these things load faster.
The Second Life Maintenance Viewer version 220.127.116.111655 has serious crash problems, a fixes have been found and may be rushed out next week (9).
This version of the viewer is referred to as the Links Viewer. No one at the TPD meeting knew why.
There is a Bear Viewer, another maintenance release, which is next in line and not yet in QA.
The Second Life Quick Graphics Viewer version 18.104.22.1681103 has a couple of things they think will be fixed and out next week (9). I’m unclear whether these are bug-like problems or they just aren’t getting complexity values they feel are representative enough to be using.
The Second Life Project Bento Viewer version 22.214.171.1240099 is going to get a new skeleton definition. I missed the Bento meeting for week 8 so, I’m behind. However, Oz Linden thinks this will be the final change to the skeleton. That means we are getting close to a final release. To me it looks like a dozen or so designers have gotten really serious about the project and are working with the Lindens to get this worked out. From what I see the Lindens are doing a good job of listening to the content creators. So, if you aren’t there now or otherwise participating… don’t be whining after Project Bento is released.
The Second Life Project Oculus Rift Viewer version 126.96.36.1995296 had apparently encountered a big bug which blocked its update. That has been resolved or so they think. So, Oz thinks they may be making progress on it and we’ll see an update.
Oz does NOT expect any of these viewers to promote to the main viewer next week (9). But, he expects they will all get updates next week. It is looking like we will have three versions ready about the same time. This suggests we will see a number of main viewer updates in rapid succession.
There is a new release of SSL (Secure Sockets Layer – is a standard security technology for establishing an encrypted link between a server and a client—typically a web server (website) and a browser), the security library used to secure log in and other encrypted Internet communications).
In week 9 the Lindens will be looking at the SSL release notes to see what the risks are. If they deem it serious enough they may fold the SSL update into the main viewer ASAP. Of course that delays other work. If it isn’t that big a deal, I suppose they may roll it into the viewer then plan to promote next.
Linden 64-Bit Viewer
Work on the Linden 64-bit viewer is proceeding. Several libraries are now updated to 64-bit. The Linden build tools are being changed and infrastructure (think server setups) are being updated to handle 32 & 64 bit builds.
The Lindens still plan to drop 32-bit Mac versions of the viewer. Oz says he looked through recent stats and there were only a couple of hundred user sessions that used 32-bit Mac versions and those were all on machines that are capable of running 64-bit versions. So, there literally is no need for a 32-bit Mac client.
I’m hearing it in a couple of places. There is a new viewer shiny coming for Outfits. In the TPD meeting I learned they are calling it: the Visual Outfits Browser… Now that is an idea I like. But, we’ll have to wait to see what it is. I’m still hoping sub-folders will be added.
Jessica Lyon, Firestorm Team leader, was asking about viewer statists and explaining why the developers need them. It is fun to know how many people are using a viewer. But, it is USEFUL to know viewer crash rates.
When a new viewer is put out it helps to know whether it crashes more often or less often than a previous version. It suggests a mistake was made or a problem fixed. If the crash rate for a new version jumps up, did the crash rate jump only for ‘my’ viewer or all viewers? If the Firestorm Team sees a jump in only their crash rates and everyone else’s rates go down, the team can suspect they made a mistake and have a problem unique to their viewer.
If everyone’s crash rate goes up, the problem is likely in the Linden or some code everyone is using. They have an idea where to look.
You can see that knowing the crash rate for each brand of viewer would let developers look at the differences in each viewer to have abetter idea of which code or combination of changes is contributing to a problem or solution.
What Oz Linde used to provide developers was overall crash rates per viewer brand, omitting any breakdown by version per brand. That report and reporting system is gone. For whatever reason this specific reporting has not been replaced in the new stats system. But, Oz can query the system and get the basic stats needed.
There is the problem of whether the data being gathered in a report is representative. Oz has been through that discussion before and still has certain concerns. One being that viewers that crash can’t always transmit why they crashed. They die before they can.
With stats there is always the problem of figuring out of stats are representative. It is mostly in how they are used. Example: 99±% of convicts eat potatoes. Accurate stat. Therefore eating potatoes cause crime. Bad deduction.
Oz also has a problem releasing all the data on all the viewer brands and all the versions to all the developers. It’s a privacy thing… developers’ privacy. Would you want all the other developers to know your viewer sucks? Of course that assumes they don’t already know. But, Oz will provide the stats on the Linden viewers to the TPD’s. Plus send individuals the stats on THEIR viewer. They’ll have some basis for making deductions.
Oz is promising stats to developers by March 1…
D______m Viewer has 4,800 users… This is a copybot viewer people can use to steal content. At the 2/26 Third Party Developers (TPD) user group there was discussion about how to block old viewer versions and copybot viewers. About 16:00±
The Lindens and developers have different ideas about successful the Lindens have been in blocking copybot versions.
Oz Linden says not to expect much in the way of new features over the next few weeks. The reason is they are working on infrastructure. In this case that means the libraries and tools the Lindens use to build viewers and servers.