SL projects update week 47 (2): server, viewer, general items

Server Deployments week 47 recap

As always, please refer to the week’s forum deployment thread for the latest news and updates.

  • Tuesday November 19th: the Main channel received the maintenance package previously deployed to BlueSteel and LeTigre in week 46. This package comprises further infrastructure changes for the yet-to-be-announced Experience Keys (experience tools) project
  • Wednesday November 20th:
    • Magnum remained on the same maintenance project as deployed to it in week 47, but which features a further update to the grey goo fence, now only applies to objects which are both large and physical. This alteration is in response to BUG-4448, wherein it was reported that building rezzers were running up against the fence when attempting to rez complex builds
    • BlueSteel and leTigre received a new maintenance package comprising those changes deployed to Magnum in week 46, with additional bug fixes.
Server Beta Meeting (stock)
Server Beta Meeting (stock)

Notes on the Deployments

Commenting on the BlueSteel / LeTigre updates during the Server Beta meeting on Thursday November 21st, Maestro Linden underlined the last four bug fixes listed in the release notes, together with the final “new feature” item, as being newly introduced with the deployment.

With regards to the BlueSteel  / LeTigre fix for  “Vehicles containing a mesh are returned to the owner upon region crossing when destination parcel is full”, he added, “I believe the issues with region crossing on vehicles due to ‘parcel full’ should be fixed, though there’s still that bug about certain vehicles sometimes going crazy upon region crossing.”

A question was asked if the change to the grey goo fence (BUG-4448) might impact llGiveInventory object-to-object transfers. Apparently there were reports n the Advanced Scripters of Second Life that people were encountering a “give inventory failure: grey goo fence: rapid or recursive inventory transfer” warning on Magnum regions following the update.

Commenting on this, Maestro Linden said, “I checked with Simon about this one; the GGF change should only affect rezzing of objects; when you simply pass items around with llGiveInvenotry(), the object geometry data hasn’t been loaded, so it wouldn’t have the opportunity to apply a penalty. However, he thinks that if you hit the GGF for rezzing objects , the GGF may also prevent llGiveInventory() from operating.”

Maestro also indicated that new restart scripts were using during the RC deployment which displayed the restart messages in “big, bold letters”. Whether these changes are in addition to the restart message changes made by Simon Linden and deployed in September is unclear; Maestro also referenced the fact that restarts will now occur as soon as the last avatar departs a region, rather than waiting for the countdown to complete, and this was a change initially deployed in September.

Week 48 (Week Commencing Monday November 25th)

A reminder that there are no deployments / rolling restarts planned for week 48 due to it being Thanksgiving week in the United States (which also means the Server Beta meeting will not be taking place.

SL Viewer

The SL release viewer was updated on Thursday November 21st to version (dated November 15)  – formerly the GPU table updates RC. This release contained no functional changes to the viewer, but saw support added for the following GPU families:  newer nVidia GTX 700 series; AMD R7/R9; Intel Iris Pro  (download page, release notes).

Fitted Mesh Project Viewer

Redpoly Inventor, who first looked into the use of custom bones for mesh garment deformation in SL

Thursday November 21st also saw the release of the Fitted Mesh project viewer This viewer includes a new avatar skeleton with additional collision bones which allow mesh garments rigged to the collision bone structure to adjust with changes to an avatar’s shape using the Edit Shape sliders.

The system is modelled at the approach first mooted during the Closed Mesh Beta and later prototyped by RedPoly Inventor, and which has been subsequently employed in approaches such as Redgrave’s “Liquid Mesh” range of garments.

The official blog post on the release can be read here, and I was fortunate enough to be given preview access to the viewer a little ahead of the launch, and my own overview is also available.

This approach to fitting mesh garments is still at a project status, and those trying it are requested to file any issues they have via a JIRA to the Fitted Mesh project.

Other Items

Copying Large Numbers of Items to the Inventory of an Object

Many people are likely to be familiar with this issue: select  a large number of items from inventory and attempt to drop them into the contents of a prim in a single go, and part / all of the process may fail. This is a long-standing problem, and there had been something of a limit of around 42 items which could be successfully transferred into an object’s contents in a single go. However, there have apparently been renewed reports of problems, and a suggestion that the threshold for moving a large number of objects in a single go may now be around the 30 mark. Maestro Linden has updated a bug report from Dan Linden on the issue with this information in the hope it will assist in narrowing-down the possible cause.

Aircraft Region Crossing Issues

Yuzuru Jewell reported further issues with some aircraft encountering problems on attempting to cross regions. It takes the form of aircraft using mono scripts with collision detection  are failing to cross region boundaries. A similar bug has been reported for mesh vehicles – BUG-4084 is “Mesh car starts to bounce like pinball after sim crossing”, but the Lab has been having problems reproducing that issue.  One workaround that seems to prevent the problem is for the collision event to be moved to a separate script compiled as LSL2 rather than mono.

In the discussion Maestro noted this approach had also been put forward in dealing with BUG-4084, except that in that case, all scripts were being compiled as LSL2, possibly because the issue hadn’t been identified as perhaps being with collision event handlers. Maestro believes the pointer towards collision events may well  be an interesting lead and has requested that anyone able to reproduce the issue and show that specific events / functions are responsible, it will obviously make it easier to determine a resolution.