XML Upload of scripts

What do you think? Would you like to be able to do XML upload of scripts?

We seem to be mashing three issues together here.



Do you mean upload script button instead of ctrl-a, ctrl-c, alt-tab, ctrl-a, ctrl-v? Or something else? If the latter, well, you can just keyboard macro that I guess.

Yeah, well, CIL obviously would solve all problems, and probably create zillions more. Could you imagine the hackery we would do if we had direct access to CIL? Woo!

LSL is absolutely necessary and vital to SL. In a world where everyone is scripting together, a layer must exist which reduces complexity for the vm watchmen to weed out bugs and safeguard against hacks. This is what LSL is. In fact, the more complex it becomes the more dangerous it becomes.




Well, this could be solved simply by letting remote script loading change inventory contents. No need to implement what we can implement ourselves.




I already do this, and I suspect many other savvy individuals do. I have a set of LSL libraries and a simple emacs macro that generates LSL from those libraries based on the symbols I access and the includes I have.

It's the FREAKING uploading into SL that is pissing me off because I always forget that one object in the rather complex object trees I have when I change a core include file. Plus I am getting tired of pulling an object out of an inventory, changing it, and then re putting it into an inventory.



Ahhh, that was directed at me. If you're developing offline where the checking is done (cause it sure aint done in SL!), how do you move from offline to online without LSL upload?






Once more, why get SL to do this when we've already done the work? They have higher priorities. If they start working on an SDK without fixing all the zillions of bugs in SL I will grind and nash and grind and nash my teeth into sand.



Heh heh, awwwh, I thought it was funny.

I think the problem is that you probably don't spend a lot of time developing within complex object trees with offline include libraries. If you did, you might feel my pain.

In order to do real development, it seems to me that this is rapidly becoming necessary.

Having to copy around scripts to different prims and what not is very tedious, and remote script load doesn't cut it.

What do you think? Would you like to be able to do XML upload of scripts?

However, once the scripts are uploaded to a single object, it is pretty trivial to distribute them automatically (pin script, sensor scan, notecard to map scripts to objects).

What I would like most to see would be enough client SDK to be able to invoke "pretend I created a script on this object with these contents" and similar from command line or local IPC.

Pretend? Eh? Can't you do this yourself?

1. Includes, let's say you had a largish project which involved dozens of scripts and then you changed a core include library which was used by every single script (because you found a bug). You have to upload every single script seperately. Notecard will not cut it, becase they're all different scripts.

2. pin script does not effect inventory items. Am I the only one using llRezObject around here? And rez / load is not a viable way to do things.

You want to upload scripts to: a) large numbers of objects in world; and

b) large numbers of objects in inventories.

That solves A neatly enough that you can add an Upload button to emacs without having to emulate input events on the SL client.

B is a bigger problem because we can not even do it manually at the moment, though all that is needed to automate it is an llGiveMyself routine.

Refactor to allow script reuse.

But the real solution is to automate the script uploads (see A). The notepad only says which script to put in which object...

B is the easy solution.

As we have very limited interaction with unrezzed objects, this is probably a system limitation. Dealt with this mess earlier while playing with fabbers -- the user had to manually drop objects to be nested into the fabber's inventory after they were built.

I just want shell access to my account and land already. Problem solved.

You mean switch everything into Say channels? For obvious reasons, this is not doable. MsesageLinked, which isn't much better, isn't a solution as well because my objects are rarely linked.

Multiple scripts within a single object, actually. Architect your code properly and the intercommunication between components is not usually a limiting factor. One of the major reasons my compiler is taking so long is that I want interfaces on inter-script communication, to allow no-modify modules to be marketed and integrated easily.

That is quite potentially the most moronic interpretation possible based on what I said. Think about it carefully, and you will probably figure out how to refactor some includes to multiple scripts sharing one prim.

Having just spent a while updating a 3 item object I can agree with Blaze that it's a nuisance, and if he does it regularly then he's hit a painful area in SL.

But I'm afraid insisting that SL completely rework their architecture because of one person's preferences for what they make things is something I receive with less sympathy.

Proper error handling, arrays, better object-object communication, bigger memory limits for scripts would all be higher on my lists and although they may not affect every scripter I suspect they're much more common limitations than yours Blaze. Sorry.

And why XML? Wouldn't YML or SML be better? What about NTP? or NNTP? RMA? SMB? IRT? LBJ? CIA?