Second Life News 2014-5 #2

RC Viewers

Interesting Viewer version – In this version the developers are still hunting for the cause of crashes. The crash rate is still too high, but crash measurement may be erroneously high. Not having clear crash reporting leads back to the Google Breakpad work the Lindens are doing.

2014-5 Third Party Viewer Meeting

2014-5 Third Party Viewer Meeting

Few Third Party Developers have tried merging the Interest List code with their viewers. The Firestorm team is waiting for a more stable release of the code.

Fitted Mesh Viewer version – Recently this version got a refresh and is likely to get one more next week (6). The fix probably arriving Monday (2/2). A later update will likely only be a merge with whatever viewer is current main version. Then this viewer will likely be in the final RC version before release.

HTTP Viewer version – This is likely the next RC viewer to be promoted to the main release for Second Life™. You probably remember this is the version that changes how the viewer and server communicate. It should have much better download speeds. It also should require fewer connections in this place a lighter load on routers.

Maintenance Viewer version – This RC viewer is the newest RC viewer in the group. It is a collection of bug fixes. The list of fixes is long, a couple of dozen, and there are 9 known bugs that remain. See: Release Notes.

This version sees the death of the GPU Table, or at least preparatory work to remove the GPU table.

The idea removing the GPU Table is to do two things; one is to ask GPU which OpenGL version is in use on the computer. The second is to measure video memory transfer performance by copying video memory chunks around.

It’s expected that this process will probably provide better GPU identification than what the Lab gets from the GPU Table guesses use now. This release measures memory transfer speeds, the cards the GPU ID, and sends that information to Lab. Analysis of the information will show whether or not things are working as expected. If this works and they can better predict the GPU settings to use they will drop the GPU Table. Not having to maintain the GPU Table will save the work associated with keeping the table updated.

DirectX is used on the Windows side to get version information. The philosophy in OpenGL is a person is supposed to be able to ask OpenGL what it can do and never bother with a version ID. The Lab’s experience is that has not been very useful in practice. Thus the calls through DirectX to try and get version ID’s. Even OpenGL relies on DirectX to get some system information.

The current processes for getting the capabilities of a computer’s video system is on the buggy side and unreliable. So, it is this new testing process will be better than the current system or at least no worse.

If the memory transfer speed testing works turns out to be as expected then some dependencies and table updates can be removed from the viewer. An easier to maintain viewer will mean more time for other work.

Google Breakpad – Its back. The Lindens are making another try at seeing how it reports problems. The effort here is to get better error reporting for tracking down bugs.

The Sunshine Project is about to put out a Project viewer. When is unclear. This will be the one with the work on bake fail. It changes to how the inventory backend is accessed. It will have more support for AISv3. It will have more of the old unused bake-code removed.

Oz Linden is guessing the test version will appear next week. But, doesn’t know.

Too Many Candidates

Oz says they may hold up on some of the RC/Project viewer releases as having too many in testing becomes awkward for the development teams.

Status Unknown

There are lots of things I could put in this category. But, here’s one that should be of general interest:

BUG-4638Horizontal & vertical offset values for normal & specular maps still randomly revert to previous settings when building. Oz is checking to see where this fix is. This bug appeared in the first releases of the Materials enabled viewers. It has gotten better but is still around.

Steps To Reproduce

  1. Rez a cube prim and edit it.
  2. Set a normal texture – blank is fine.
  3. Set a specular texture – blank is fine.
  4. For the normal and specular textures, use the spinners to change the horizontal and vertical offset values and observe how often the values you have set do not stick.

The problem is aggravated by the use of negative offset values. Remember. An offset of -2 is the same as an offset of 8. I have yet to test it, but sticking with positive offsets should reduce the problem.

Group Bans

Oz talked to Baker, who has been scarce at user group meetings. Baker is in the later stages of his redo of Group Ban code. A viewer should be arriving soon™.

I think the delay in Group Ban is Baker being new to the SL code. Seems he thought he had done. Then when other Lindens reviewed his code they pointed out the tripwires he had missed. So, for weeks now he’s been rewriting to handle some of the intricacies of the SL code.

LEAP Motion

No Updates on Leap Motion Development. Oz hopes to have an Oculus Rift and LEAP Motion Controller. That was my hope too. I have a LEAP Motion Controller. But, my experience with the LEAP Motion Controller were not that good. There have been updates to LEAP, so I need to update and play with it again. That said, my experience so far is people are not ready for it as it is now.

What hand motions will users find ‘intuitive’? What hand motion will we use to move the avatar forward? Or to shoot a gun? Jump?

Do you really think we are all answering those questions the same way?

We still have more new people coming on the Internet every year than there are experience users. This means the majority of computer users are novices, by a huge factor. What do people think is a good signal to move an avatar forward? In RL we walk forward. There is no easy way to do that with a computer, even with mobile devices. As much as people think a more intuitive interface is possible, some things are going to be controlled by arbitrary interface instructions that people have to learn.

Remember too that Drax was playing with LEAP. See: My Second Life LEAP Motion Experiment. May be he will follow up on his experiences with LEAP. I wonder if he is still using it?

Viewer Statistics

Various reports that the Third Party Dev’s want are still in the works. I think that means we still have no idea how many people have Advanced Lighting Model (ALM) turned on.  This becomes more important as the main release of Fitted Mesh nears.

Materials and clothes just goes together. I expect clothes to be more interesting as people learn to place the details in Normal Maps. I also suspect some designers will delay adopting Normal Maps until they are sure a significant majority of users can see Normal Map effects. Hopefully, enough designers will go with the new tech to push people to more quickly enable or adopt hardware that performs well with ALM.

One thought on “Second Life News 2014-5 #2

  1. Pingback: Second Life News 2014-6 | Nalates' Things & Stuff

Leave a Reply

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