The third party viewer developers’ meeting was Friday. News from there:
A viewer hot fix was promoted to the main viewer’s channel. This was on top of the Fitted Mesh promotion earlier in the week. This ‘hot fix’ was to fix the Facebook posting problems. Since it was Valentine’s Day weekend Linden management decided this was the RC (Release Candidate) that should be promoted.
The HTTP RC Viewer is doing well. Jessica Lyon says that Firestorm’s merges of the Fitted Mesh and HTTP code are working well for them. That doesn’t mean it will work well for everyone trying to use the viewer. It just means that within the group currently testing it, things are going well.
As of Friday the Maintenance, Breakpad, and Interest List RC Viewers are in various stages of development. Their next upgrade will be to add in the Fitted Mesh and Facebook changes and retest as RC Viewers. Expect that to happen in the coming week or two.
Sunshine and Merchant Outbox project viewers are in somewhat unknown status, meaning Oz Linden doesn’t have all the details… not all viewer projects are his to deal with. Nor does he set the release priorities. To him it currently looks like the Sunshine Project is likely to go to RC status ‘pretty soon now’. Since I am translating this I’ll take that to mean this coming week (8) or next week (9) and more likely next week (8) than not.
It is less clear when the Merchant Outbox Project will make it to RC status. I’m not sure but I think the outbox thing is mostly a Linux-side issue that has been in progress for some time.
Sunshine in some sense has not been promoted to RC as the Lindens are trying to hold the list of RC viewers being sent out to people at no more than 5. Now that an RC has been promoted, Sunshine will be merged with the latest releases and we will see it appear as an RC viewer.
Why just 5? I suspect there are only so many people with ‘Willing to run RC…’ enabled. More than 5 would be making smaller tests groups. The smaller the test groups the less diversity in equipment running the RC viewer which make for a less effective testing.
Sunshine is using AISv3, which is already deployed to the servers*. We just need the viewer side changes to take advantage of those new changes. – *There is some confusion on where AISv3 is and is not deployed. Looking back though this blog I don’t find where I’ve written about the release of AISv3 to the main grid in the last 3 months. But, I remember writing about it. Nyx and Oz think the change is grid-wide on AGNI, but want to check before saying that is true. I think the same but can’t find the notes today.
Vivox is working on updating their systems. They got new software that reports bugs and surprise they have plenty. So, they are suddenly into getting them fixed.
So, in looking at their logs they find Second Life people are using a number of older versions. Since the Lab has updated their viewers and servers to use the newer, if not newest, versions that leaves third party viewers and those people resisting updating their viewer.
Vivox says if third party developers will tell them why they are not switching to updated versions, they will fix whatever problems Vivox may have that are in the developers’ way.
Jessica points out that Vivox has not been good about responding to requests to fix things/exploits. Oz points out that Vivox came to ‘us’ asking. So, he suggests we give them actionable information so they have a chance to fix things.
Vivox will have a new version out, probably within weeks. This new version will have fixes and improved voice quality. The Lab plans to jump on the update.
Firestorm is using the current ‘new’ Vivox (4.5?) in their beta viewer (FS: 4.5.1). They have some problems with it. If you have access to JIRA BUGS see BUG-3616 – Grid wide voice stutter issue. Alexa Linden points out, “…Unfortunately voice outages/oddities are very hard to triage after the fact.” Meaning they need to catch the problem in the act.
Oz told the group that he was employed to work on Voice Over IP (VoIP) for some years. He cautions developers to be careful about blaming a specific version for a problem. Trying a version, having the problem, swapping versions and not having the problem doesn’t really narrow down the problem as one might think. Consider: a number of things that affect VoIP have changed in such a test; the viewer has restarted, time has passed, network load in your leg of network probably has changed (your horny neighbor completed their porn download), the lady downstairs finished her trashy NetFlix movie, and you refreshed/changed your network connection by relogging. Also, the neighbor’s microwave may have finished cooking dinner, shut off, and be allowing your wi-fi to work properly again… well, stopped interfering at least. Oz’s point is there was far more happening than a simple Vivox version change.
Some Firestorm team members have seen the problem in Firestorm Beta and the SL Viewer (3.6.10-283403 Nov 6 201). Whirly Fizzle and Ed Merryman did some testing with pre-Viviox updare and post Vivox update viewers. You can read about that in BUG-3616 Comments.
Ansariel Hiller, of the Firestorm Dev Team, has found a work-around. Expect it in the next Firestorm release.
BUG-4573 – Avatar Lip Movement audio feed is now taken post the Voice mixer fader and not pre. This is undesirable. This bug appeared with the Vivox updates. It is a problem for machinimists and furries. No one has found a fix or work-around AFAIK.
Oz says the Lindens will be providing a 64-bit version of the Vivox library for use with SL.
Oculus Rift Viewer
Oz says right now the Oculus Rift team keeps saying, just one more bug. Eventually that will be true and we will see a project or RC version of viewer with Rift support. There is news on this subject from Firestorm. See the next article on the Firestorm meeting. I have it mostly written and am proofing it next.
Shadows and Fitted Mesh
BUG-5056 – Shadows cast by rigged alpha surfaces ignore alpha completely. I’m unsure how many people actually notice this problem. I hadn’t.
Some viewers have the option to turn on Look-At… beacons, crosshairs, indicators… whatever you want to call them. They show where an avatar is looking. This is a debugging feature the Lindens built in for working with avatar animation long ago. Some viewers also have the option to label those indicators with an avatar name so you know who is looking.
Niran is one that thinks the feature should be removed as it causes drama. Oz points out that as best he can tell all of SL is drama causing. I agree with him. In general life is drama causing.
Niran thinks avatar names should be removed and that that would reduce drama…
Jessica points out that in SL the Look-At with labeling is used by teachers to make sure students are looking at the correct ‘thing’, whatever that may be. She further points out that this feature has been in SL for years and received few complaints.
There is a history between Firestorm and Niran. Check the video at time mark 28:00 +/- for that drama. But, it is my opinion there is history between Niran and everyone… which should give people some idea of what is going on.
Several things were pointed out in the discussion. First Look-At markers are a key part of the avatar animation used in SL that keeps the avatars from looking like zombies, lifeless. The chance of removing that part of the animation system is nil.
Oz points out that there are numerous and various features in third party viewers (TPV) that are included just because someone figured out how to do something. Often there is no visible reason for the feature being included, no useful purpose served by the feature, and often the features provides information that is misleading or poorly understood by users, if at all understood. But, he does not see those reasons, or lack of reasons, as grounds for prohibiting or forcing removal of those features in TPV’s.
More to the point Nyx Linden asks, “Do you want Linden Lab policing exactly what information gets visualized in third party viewers to that extent?” I know my answer is no, I don’t want that. I prefer freedom.
For me this is an attempt to ‘legislate’ moral human behavior and force people to behave as someone thinks they should. With several thousands year of history behind us we know that doesn’t work. Drama is part of life, SL and RL. It is up to us to learn how to handle it. Creating more laws creates more problems and generally reduces freedom. At one time the laws against gay sexual behavior were consider appropriate. Now they are an obstruction and conflict with modern thinking and personal freedoms. Yet in many places we are stuck with those laws… ever hear of Putin?
People are always trying to create more restrictive systems to have things the way they think they should be. Fortunately Oz thinks TPV Dev’s should be free to create and develop within the system we have. Only when features become a significant problem, meaning for way more than just a couple of people, should the Lab consider restrictions. This is like the original American concept that states would have rights to experiment with government and thus our nation could adopt and reject policies based on what we saw actually working. Oz’s idea is what works will get adopted into all viewers.
There are some problems and lots of confusion right now. There is a known problem with mesh upload not handling bounds limits correctly. It is going to be ‘some time’ before the Lab gets to where they can work on it. There is a problem I mentioned earlier in Blender 2.69 that is messing up Fitted Mesh uploads. More JIRA complaints will likely change the priority on that problem.
This code has moved back into QA and will likely show up in an RC viewer soon. It is likely to suffer the same problems SunShine has run into, too many release candidates in cue. That will delay it reaching RC and promotion to main. But, at least it is on a release bound path again.
I am guessing that puts it 4 to 8 weeks away. In that time we will likely see a test in ADITI.
For a time we have been short news of what is happening in SL. Little server news is getting out. It seems part of the reason is the viewer cue is backing up. And that is in spite of the new RC process. These backups mean the server changes can not be fully used. With so much in cue, the Lab can spend more time getting other server projects completed.
After the features in cue are out, I suspect we will start to hear of the new stuff on the server side.