Mono on the beta grid

Mono is comming:



http://wiki.secondlife.com/wiki/Mono

/me confesses to reading this and thinking "yeah.. heard that before.. that article is probably a year old."

It's actually only a week old. And yes, it was written by a Linden.

/me starts to drool.

Hedgehogs drool?

Scripting ones do, when they hear that Mono may actually be happening soon..

Well, this seems like great news. But I doubt that users will see any direct impact for quite a long time.

For starters, Mono will make scripts run much faster. Previously, LL was throwing around numbers like 150 times faster.

From a practical view point, this means that the sim server will not have to work as hard to run all the scripts in a sim, so it will have more time to do other stuff, so sim server related lag will be decreased.

It might also mean that programmers won't have to care as much about runtime efficiency, so instead of hand tuning a script, they will go ahead and write something else that is needed.

Script that contain busy loops will waste time much faster than ever befre.

It is still a step in the right direction Lee, I drool along with Meade (and no I'm not a Hedgehog just a resource hog). Let them get it in and stable, then we can start pushing for refinements to LSL or, even better, C# etc.

that one line reminded me of a game I tried to play... it was old written for a 386 machine... I loaded it onto a pumped up 486 (roughly p90) and made the mistake of blinking... poof game over... it can be dangerous to time things to processor speed... makes me wonder if they'll modify those delay times

I'm less worried about execution time rather than script memory. And please for the love of god include the #INCLUDE statement. Please. K. Thx. =^_^=

~TENK

phht so, they do that all the time so get a bucket

So if we want to program in Mono instead of LSL, what will we need?

This is a dream many have had for awhile now 2005. It was only yesterday it feels that mono was talked about. But we are now very close to real time on the main game. slowly we are stepping closer to this lighter less blucky scripting languge. The sooner the better.

Nothing. Mono will first simply run things (better) behind the scenes. Eventually it might enable us to program in other languages like Java, C++, C#, etc., but not if the Lindens have anything to say about it.

Is other terms faster, clearer, smaller scripts that will weight less on servers land (simulator). Were once a script was over populated with LSL that lagged a island (simulator) now will be half as small with less commands but sleeker running time. With less lag for those that populate the island (simulator). More agents on a island(simulator) with less scripts running as there would be on todays standard. Mono Mean ( LESS IS MORE ) In better terms where there was 7000 scripts were running........But after Mono was introduced say 3000 would be instead. All inturn means a faster sleeker less blucky more agents on a simulator more 200 up to 500 agents we are possibly talking about ).

Usagi

I don't have a problem with them sticking to LSL, as long as they stop treating it like the bastard cousin that everybody keeps trying to hide under the porch. There's some llStuff that would be really nice to get and there are some language features (/me screams "switch/case" for the bazillionth time) that would make scripting in LSL a _lot_ easier..

I've got lots of experience in the C trilogy and have done my share of Java. I don't see these languages as really fitting with the SL model, though - SL really wants something that's event driven instead, IMO, and LSL is exactly that.

Hopefully some of the processor time saved by using Mono can be used to better throttle scripts. I still intend to fine-tune my scripts to be as efficient as possible, as I don't like to think that my products could be using a lot more processing time than they need to. Is why I'm trying to find a way to redo a very complex HUD which currently requires one script per button (it has up to 48! EEP!) and do it in one script with a few helper scripts that can be put to sleep. But it's no small task.

On that note, hopefully the overhead of idle scripts (just sitting there waiting for an event) is greatly reduced too.

I have a great fear that it may be a looooooong time before LL will up the memory constraints on scripts even after MONO is implemented. More memory equals more wear and tear on the backend. All of us have seen some truly horrible scripts floating around and they must be just the tip of the iceberg. I could never disagree with allowing everyone to script because that is how I learned, but unfortunately there is a price to pay for that luxury.

Hopefully I am wrong.

That's an easy quick fix, just name each button and use the detected functions to determine which button was pressed. I converted a 10 script HUD into a 1 script one once, very big improvement.

Timely and relevant. Part of a reply from Kelly Linden to the Scripters mailing list:

"LSL is unlikely to fundamentally change ever, this includes new
datatypes or other language features such as include/import etc. The
mono project is underway and making great progress - while I'm not sure
of its official status I have played around on some sims with phase 1 of
mono. To elaborate on the mono project there are multiple phases, and
the first phase is not going to help you in this regard. The first
phase is running LSL as it exists today on the mono VM - this gets us a
huge speed improvement and better memory/resource control but nothing in
the way of language features. The next phase is to allow scripts to be
written in a standard language that compiles to mono - probably C# but I
don't think that is written in stone. This phase will get you the
language features you need, but it is a long way off. I'm sorry but I
don't have a better ETA than that. The final phase we all hope for is
to have tools available for third parties to create the libs and hooks
needed to create scripts for SL in any language that can compile to
mono. These "phases" are really just the biggest milestones that I know
of and there may be "smaller" phases in between."

Every button can be in one of 15+ states and perform an action on any channel the user wishes. It's intended as a scriptable HUD which reads button style and actions from a notecard. So not so simple unfortunately

The latest build of a Mono-enabled simulator is looking good. There are still some bugs, but so far none major enough to prevent a beta. Unless QA finds a showstopper our plan is to begin a Mono beta program next week.

This will be small to start with -- four sims on the beta grid will run Mono-enabled. The rest will still be running the Havok4 beta. We'll make a viewer available which will connect you to the beta grid and will contain the Mono UI (which is really just a "Compile to Mono" checkbox).

Once on the beta grid the Mono team (Babbage, Scouse, and I) will hold office hours there to answer any questions you have on current or future plans. Please see https://wiki.secondlife.com/wiki/Mono for more information. That wiki page is for the general resident community, so it contains some basic info about LSL that you may want to skip over. The page will be updated with Mono status throughout the beta.

Cheers,
--Peri

That would be very nice. It used to be that way a long time ago, IIRC, and then they changed the scheduler, and idle scripts started taking up CPU cycles. I have some old products that are fairly clean script-wise (IMO ), with every major asynchronous function pulled out into its own script. Today, I'd probably try to make that much more monolithic, or throw in complex sleep/wakeup synchronization, to avoid the idle scripts. Which would probably result in more actual code being executed, just to avoid having idle scripts. It would be very nice if I could go back to more 'multi-threaded' designs without worrying about the cost of idle scripts.

Ha!
We shall see how my scripts will perform on Mono given that they can force a class 5 sim to run scripts at up to 4.2 million instructions per second...

::::::::::::::::::::::::::::::::::::::::::::::::::::::::thud

for those that missed it, LL is using island owners as voluntary guinea pigs for testing H4 on the main grid, there's an 'early adopter program' available to sim owners, letting them choose to run the new H4 code.... hopefully they do this for mono too =)

I thought the script compiler was going to be different and that an old script would pretty much behave as (slowly as) it used to, even if run on a MONO-enabled sim. If that's true, I'm not sure the het grid stuff will help on this, unless they had both compilers in the viewer.

Do we have details on how all this stuff will be organised? Thought I also heard something about distributed compiling, too..