Jonathan wrote:
Hello.

I wanted to throw out my interest to gain like contacts in SL/libSL developments. Specifically at this time, the LSL compiler to LSO and CIL. I heard someone worked on it, but I haven't gotten their name.

We had a developer that went by asterick in IRC that was working on an LSL to LSO compiler. He was intending to market it as an optimizing compiler for people who wanted to push the limits of their scripts by using smart compiler optimizations to save memory. It was going to be a closed source product at first until he recouped some of the costs, and later put it in the libsecondlife repository. He finished documenting the LSO format on our wiki, and apparently had gotten over halfway done with the compiler (it was compiling simple scripts and uploading and activating them on the grid already) before real life took over. The last time I heard back he was hoping to continue work on the project, but said if he couldn't find any time he'd release his current work so others could take over.

While LL does have programs to compile LSL likewise, LL has shown in recent roundtables to extend to the hand to entrepreneurs to create a development environment external to the SL client. LibSL is useful to communicate between such an external development environment and SL servers.

The LSL compiler is a more immediate project. The CIL relation is practical and implementable now, but it is still visionary. As you ask what does it have to do with it, that depends on how far down the rabbit hole you want to go. For example, there is some talk on this list about sim server development. A VM besides Mono's or .NET's may be desirable.

The SL servers only run LSO bytecode right now, and LSL compiling will be moved to the servers much sooner than they will transition the grid to a mono scripting backend. An LSL to CIL compiler would be interesting for a local sandbox sim project, but wouldn't be useful on the grid for a long long (long) time.


On Babbage's blog, it was stated that in was not easy to encapsulate a CIL program's state into a packet to transfer to another sim as objects move from sim to sim. That would explain why only one SL sim runs Mono at this time. The VM I have may demostrate a solution to that. Otherwise, it may provide the potential to host a local sim. These are a few ideas.


Ok, I just want to clarify that none of the libsecondlife developers work for Linden Labs, you probably want to talk to them directly about feature suggestions. As far as a local sandbox sim, it would have no need to pack up it's state since there would be no other sims to transfer scripts to. If you are talking about building a separate grid that's a bit out of the scope of libsecondlife and more in the realm of the Open Source Metaverse Project, Croquet, or other endeavors.

If I see asterick around again I'll direct him to this thread and see if I can convince him to help the compiler project move forward. There's a lot of possibilities if an open source compiler existed (think Eclipse and Second Life running side by side).

John

_______________________________________________
libsecondlife-dev mailing list
[email protected]
https://mail.gna.org/listinfo/libsecondlife-dev
http://www.libsecondlife.org/

Reply via email to