Well… let’s start with the JIRA Change. In the Content & Mesh User Group meeting Nyx, Vir, Hoz, Oz, and Alexa Linden were all there. I suppose they were there to handle JIRA Change blow back. It wasn’t that bad.
The first topic of the meeting was the JIRA.
“Given that the jira is no longer going to be an effective means of accurate two-way exchange of information, how can we communicate with other users about problems with the content creation processes and their solutions, especially in the context of this user group where the commonest question is “can you give me the jira for that issue”? Drongle McMahon
Nyx Linden responded, “Good question, this is one we’ve been getting a lot. Collaborating on an issue is definitely an area of concern with the new JIRA setup.
For the time being at least, let’s try collaborating on the details of a bug report in the forums and taking the results into a JIRA. We can also discuss details of a particular bug report here to make sure we get the details straightened out before filing an issue.”
Having read that, remember who Nyx is talking to. The UG meeting is a group of people very much into SL and willing to spend time with Lindens in UG meetings. So, Nyx is not expecting everyone in SL to attend a meeting and talk over a bug before filing a JIRA. But it is a possibility for those of us that do attend UG meetings.
Later Nyx commented explicitly, “Not every bug needs to be discussed here; this is just a good place to have those discussions, if you want to work with other people on filing a bug.”
In some cases people think bugs will be discussed in the wiki. Personally I’m not sure that works. With the LSL problems that may be a possibility. Each function is on its own wiki page and has a dedicated Discussion Page. I’m not sure how many people actually use them or even know they can. Then there is the matter of whether people know they can edit the wiki. I’ve personally made a number of changes to the wiki and added information and examples. I’ve even discussed problems on the Discussion pages. But, I am likely in a very small minority of SL users.
Not Just Lindens
I filed BUG-9. It was not long (hours) until a Linden responded to it. But, it was not a particularly satisfying response. I realized what other facts I should have added to the initial report. But, initially there was no way to respond. Off went an email asking how I could follow up. I may have over looked the Comment Button at the end of the page, but I am pretty sure it was not there prior to the email.
Alexa Linden responded shortly after the email. Sometime after that Lance C. commented on my BUG-9. Now how did he see that bug report? Today I found out.
It seems active contributors, support helpers, and a few others can see all the BUG Project issues. It seems there are groups in the JIRA. If you are and active and have a signed Contributor Agreement filed, then you’re in a JIRA group that can see the BUG issues. Also, if you have been designated a Helper, then you can see the BUG issues. So, there are a few more people than just the Lindens that can see BUG issues and comment on them.
I gather from talking with the Lindens that some consideration is going into how to handle allowing more visibility.
Since it is not possible for most to see BUG items or discuss them or even comment on them a suggestion has been made that discussion occur in the forum. Just like the SLUniverse’s duplicate BUG report list I don’t have much hope for it. How many of us are going to fill in the JIRA and file a duplicate in the forum or on SLU? Not many is my thinking.
I asked if it would be possible to have the JIRA push a copy of the reports to a special section for the forum as they are filed. We could then search for bugs in the forum. We would also be able to discuss them.
This would also place the discussion where it could be moderated by the Forum moderators.
Alexa Linden says they are looking at doing just that. This does not mean it will happen. But, they are thinking along that line.
Not much time has passed, but it seems the decision to change the JIRA was made at high level and the Lindens we meet are not going to be able to override it. Follow orders or follow the unemployment lines is basic corporate motivation… not that that is the threat here. But, it is common work place thinking.
The staff is looking for ways to make this work. There is no doubt the work-a-day staff realize where help was coming from. They are striving to figure out how to keep that help and separate out the wackadoodles.
It looks more and more like we are going to live with the change for a time before any big change to the current system is made.
We have not heard much on this. But, work is proceeding. This is the project that is attempting to resolve avatar bake fail issues, which is what has happened when an avatar fails to rez remaining a particle ball, grey, or clothes are fuzzy. The planned fix is to create and avatar appearance baking and caching service for SL. Then the current bake process will be removed from the viewer.
Nyx reports, “We’re doing some pretty significant moving around of code in the viewer to get ready for server side texture baking and we could use a hand. We need to have a set of test avatars to make sure that we haven’t broken anything, so I figured there might be some creators here who would have some complex avatars that would be good for testing. I’m especially interested in making sure we do the right thing with outfits that use multiple clothing layers, flared pants, alpha masked avatars, and the like.”
So, what is meant by a complex avatar? Well, it has nothing to do with attachments; mesh, prim, or sculpty. It does mean and avatar wearing multiple levels or layers of clothes and alpha layers.
How does one give an avatar to a Linden? In this case, one gives the Linden an Avatar name and an Outfit name. By Outfit I mean a new Viewer 3 Outfit. The Lindens will be able to find the Outfit and run tests with it. One little consideration… it needs to be an Outfit you have on ADITI.
Changing your password updates your ADITI inventory by copying it from AGNI. There are still people reporting problems with inventory after changing their SL password. That is bug that is fixed but has some lingering affect. I STRONGLY suggest you wait 24 to 48 hours after changing your password to log into ADITI. Also, I advise you not to make the change on a Tuesday or Wednesday when system roll outs are occurring. I’m not sure whether it makes a difference or not. Whatever, if your inventory is Borked, change the password again and wait. Repeat until it works.
Nyx prefers email, but Note Cards work too. IM may work, but may cap out. Send and avatar and outfit name and mention avatar bake testing.
This project change is still being made. Possibly this month it will make it to the preview grid, but maybe not. If there are problems, it will get delayed. There will also be a project viewer released when the service is ready for testing.
Once testing is in progress you can send a picture of how an outfit looks in the ‘standard’ viewer with the avatar bake feature. Along with it you can send a picture showing a problem that occurs in the project viewer using the bake service. Including the avatar and outfit name would help.
Mesh Physics Limit
I did not know about this…
“After spending days on a project, I found that a custom physics shape only kicks in [with objects larger than after] 0.500m, below that it [uses a box for the physics shape boxes up]. On a prim [the physics does not change to a box until the prim size goes it boxes up] under 0.065. Is there is a reason for this? And will this remain at such a large size for a custom [mesh form] for the future? Obviously this causes issues and inhibits smaller scale builds when smaller and more accurate physics are required.” Sheena Kamala
To understand the problem, imagine a table top maze that you want to roll marbles through. The marbles are prim spheres. But, to save prims, the maze is created with mesh. The walls of the maze will be thinner than 0.5m. However the physics engine will change from the physics shape you provided to the default cube shape. Your marbles will never make it through the maze.
Nyx says, “There is indeed a reason: the physics computations get significantly more complicated the more triangles you pack into a given space, so when you shrink a model down it gets more and more expensive to compute. Prims have less of an issue because their shape is determined by their parameters, not just a group of triangles.”
I take it to mean the physics engine can decide how many triangles make up a prims physics model. With a mesh item the physics model created or provided at upload makes for a fixed number of triangles. This is why scaling things down can drive up the Linden Impact cost. While other costs go down due to scaling smaller the physics cost is going up.
Nyx explains, “We set 0.5m as the size when meshes switch to a box physics shape, the size of the switch is fixed, not variable.
The switch happens at run-time on the server side. Triangle counts are not irrelevant, but the limit is not based on triangle counts. Triangle counts still have a performance impact.”
There are still people that think Pathfinding is having significant impact on region performance. If you have a region that is being lagged by having Pathfinding enabled, get the region name to Nyx Linden and they will check it out.
Pathfinding was designed to degrade the Performance of Pathfinding characters when the load got too heavy. I should NOT slow down the region.
Nyx Linden said, “There should be little to no performance impact with enabling Pathfinding. If you put a bunch of Pathfinding objects on your sim, the quality of the Pathfinding should degrade instead of the sim’s performance.
If anyone knows of a particular region that can reproduce worse performance with Pathfinding enabled, please let us know and we’ll investigate.”
I asked Gaia Clary if she has any news on when we might see Blender 2.64. The rumor is the end of September.
Does anybody know whether the 0.5m limit for the physics shape is on *each* side or on *one* side?
Not me… I would have to test it to see if just one dimension triggers it or not.
I find it really interesting that the same reasons Nyx offers as an explanation on mesh physics in relation to small size are the same exact reasons given when asked why mesh items start counting exponentially more land impact as an item is scaled up to larger sizes also, like trees. This has been blamed on physics calculations by both Oz and Nyx at different times. So…which is it? the calculations go up when something is small because of physics, or the calculations go up when something is large because of physics? If it is both, which wouldnt surprise me based on the multi-page explanations ive seen of just how land impact is calculated (Rube Goldberg comes to mind), can someone please explain to me how this is possible in simpler terms? At this point, getting the same answer to both questions makes it sound like the mesh meetings of old, back during the NDA days, when corporate non-answers were the standard reply for just about anything that was asked or requested.
As mesh scales up the size of physics collision triangles increases and the physics calc cost goes down. But, the render cost goes up as visibility increases and higher LoD is held. The ideal Land Impact is when a model is not scaled and is designed to be its final size without scaling.
I have just tested this on server 264357 (BS 12.09.04) with a hollow cylinder mesh with a decomposed physics shape, following physics shape changes with the physics shape metadata display. For this mesh, it doesn’t appear to involve 0.5m at all. What happens is : (1) if all three dimensions are less than 0.2m, it uses the bounding box. (2) otherwise, if any two dimensions are less than 0.2m, it uses the convex hull. (3) conversely, if any two dimensions are greater than 0.2m, it uses the supplied physics shape. No differences whether it’s physical or not. (Note – the physics shape display updating is unreliable when the object is not selected)
Thanks!!! I appreciate all the testing you have done and published.
I did some quick tests, but nowhere as detailed as that, and I couldn’t see anything special about 0.5m either. The behavior Drongle describes there looks consistent with what I saw. Thanks for the detailed work.
Pingback: DOH! Point gun at foot, pull trigger. [JIRA CHANGES] - Page 9 - SLUniverse Forums
I script, but cannot attend meetings inworld. So I’m dependent on searching the Jira (and the occasional set of meeting minutes) when confronted with a seeming bug. Seeing a precise description of the problem, and others’ experiences of how they encounter it, helps me avoid or workaround bugs, most of which I never post comments on. Some people will probably react to this new blindness by filing a lot of bugs; I’m rather the oppositie – LL’s seeming disinterest in allowing scripters to be productive will simply discourage me not only from filing bugs, but also from using newer LSL and server features. So please, any encouragement to the Lindens to allow visibility to all residents, as well as comments/workarounds from active contributors, will be most appreciated!
Wikis are an extremely clumsy platform for discussions. The forums would be a better choice. But in addition to that, allowing a responsible subset of users to contribute helpful info directly on the bug, and allowing all non-banned users to see it (even when LL has begun working on it), is vital.
I agree! I don’t use the same process as you and I do post in the wiki, my process is similar. I use the search JIRA on function names often.