Tuesday a new version of the server rolled to the main grid. Only one high priority fix made it to the main channel this week. This version contains the fix for SVC-7283 – llRemoteLoadScriptPin and llGiveInventory fail when installing non-full perm items into worn attachments.
This problem was causing the permanent loss of scripts in worn objects. As it was breaking existing content, this was a top priority to get fixed. It was considered enough of a priority that it was placed on all three Release Channels to assure we would have a fix this week, provided the fix passed testing, it did. See Second Life Server News Week 38 for more information.
Some are upset that it took 2 weeks to roll out this fix. The plan is to test and test and be sure a fix actually fixes the problem without breaking something else. That means a week of testing in ADITI and a week of testing in a release channel. The Lindens thought this fix important enough to place it on all 3 RC’s. If it had failed it would have failed all the release channels and delayed all other fixes in progress. Their concern and urgency on this issue pushed them to use all 3 channels. But, for some that is not good enough. That’s life.
The Fast Scripts Project rolled to this channel. There are functions that make no change to any object properties that have throttles, a delay before they can be called again. The Lindens are removing those delays from most GET functions.
Any benchmarks you are using should run faster. You can find benchmarks here: Category:LSL Benchmark. Most of the benchmarks are math heavy, so the amount of improvement is partially hidden. People will see the most improvement in Mono scripts that access object parameters. Scripts using the function llGetPos() will generally see the most improvement.
LSL2 scripts will also see some improvement, but very little. Now that the Mono Script Stalls Sim problem is mostly fixed one will see more improvement by changing to Mono.
The more scripts in a simulator (I suppose the Lindens use the word simulator to mean multiple regions in the same simulator instance or server) the more improvement one will see.
Test benchmarks in the main channel/grid and then test them in Blue Steel.
There is a security problem in this upgrade. LL policy is they won’t talk about it. I don’t have access to the JIRA item, so all we know is that there is some problem in this upgrade. So, it may not make it to the main grid/channel nest week.
Another Server Maintenance upgrade is on the Le Tigre release channel. Several things made it into this update. See Release Notes/Second Life RC LeTigre/11.
This new function is enabled in this channel. Ray cast functions are used in many games for weapons. With the nature of SL it is often better to use objects as projectiles to move the calculations into the highly optimized physics engine and detect collisions rather than calculate the collision in LSL/Mono. The function will hopefully replace Sensors in some situations.
If you find a cool application for the function let the Lindens know. Falcon, Kelly, and Oskar would like to know.
The update includes the new parameters:
NOTICE: The above links are not working as I write this. At some point they will work as the LSL entries in the wiki catch up.
These are parameters that can be used with functions like: llSetPrimitiveParams.
The function llSetPhysicsMaterial( integer material_bits, float gravity_multiplier, float restitution, float friction, float density ) is on Le Tigre. The converse function llGetPhysicsMaterial() is also available.
A number of bug fixes are also included. Several are fixes for issues with llCastRay(). You can see the list in the release notes.
Also, the freeze on select problem is fixed in this release.
As of the Thursday meeting it looks like this update will roll to the main channel next week. That could, of course, change between now and then.
Very little changed on this channel. The Lindens are working with permission issues. Changes and fixes have been made to the permission system and server settings. Users should see little if any affect on existing content.
The ADITI grid is getting new Operating System software this week, or as the Lindens call it a kernel update. See [Upgrades to ADITI (Beta) Grid Commencing Thursday, 29 September. This is a response to the TimeWarp issue. This is the bug that stops the simulator and stalls all the regions on it for 30 to 60 seconds. That generally logs off any avatars in the regions. By the time they can relog, the problem has self corrected, which makes for lots of frustration.
The ADITI kernel upgrades are an alpha test of the TimeWarp fix. In week 40 we should see this upgrade move to the Release Channels. This is going to be a somewhat painful upgrade. Because of that the upgrades will be made in off hours meaning… off hours for SLT/PDT.
The same will be true for the release channels in week 40 and the main channel in week 41, if things go well. If things go well we will have kernel updates across the grid by mid week 41.
Homestead and region balancing in the Release Channels will happen in week 40. For a recent test of a scripting upgrade most of the homesteads in the RC’s were moved into the Magnum channel. That is going to be corrected and they will be distributed evenly among the RC’s.
All this is happening in addition to the regular server software upgrades. Rolling out new simulator software is automated and a regular process that has been polished to work well. The allocation of regions to the various RC’s is not an automated process. Neither is the kernel upgrade of the servers’ operating systems. So, people will likely see two or more restarts as all these changes are made. The re-allocation and OS kernel upgrades will happen in off hours and span over several days.
Expect lots of restarts in RC’s in week 40 and in the main channel in week 41.
It is hoped this is a complete fix for the TimeWarp problem.
Havok 2011 Update
Falcon Linden sent word that SL is going to upgrade to Havok 2011.2 for the physics engine. The advantages are mostly in the Havok Math Library. We’ll have better speed and precision. This upgrade will also make future Havok upgrades easier.
Falcon will NOT be tweaking collision tolerance or anything this time, so there should be fewer instances of content breakage. Well… we can hope.
Summing It Up
Seems we are still in a maintenance phase which is mostly fixing things and improving performance. We are also getting script functions to access and control parts of SL associated with mesh and ‘gamification’ pathway stuff.
Brace for lots of restarts in the RC’s next week and the week after in the main grid.
It would have been nice if they had applied the test and retest philosophy to the rush job they put through that caused this problem. It was apparently an urgency to stop content theft that led to the original bad “fix”.
Security fixes are a special case. Even then they run them through the same QA process unless it is a high risk exploit. The one that lead to starving Meeroos was in testing for more than two weeks.
Fixes are seldom perfect and often have unintended consequences. The consequences are often not apparent until massive numbers of people are using the software. That is the nature of software development.