tools Is there a way to do TDD?

[primerib1]

As the title asks: How to do Test-Driven Development w.r.t. LSL Scripting.

Likely will use a combination of external tools. And need a 'pipeline' to work, a 'toolchain' so to speak.

So feel free to share / discuss.

Edited January 21, 2023 by primerib1

[animats]

What do you want to test?

LSL programs are not that interesting in isolation. They have to interact with the world to be useful.

That said, on one occasion I really did need a serious test tool. I wrote a maze solver for my NPCs. I had to do that by developing it in Python, debugging it there using a random maze generator, and then translating the result to LSL.

[primerib1]

38 minutes ago, animats said:

LSL programs are not that interesting in isolation. They have to interact with the world to be useful.

And that's exactly what I want to test.

Like:

• Test if a toucher is blacklisted => should return an error message to the toucher
• Test if a toucher's active group is blacklisted => should return an error message to the toucher
• Test if a toucher is not blacklisted but no items to give from inventory => should return an information message
• Test if a toucher is not blacklisted and there are items to give => should receive the items

Things like that.

So that in the future as I add features / fix bugs / introduce optimizations, the externally-visible behavior of the script stays the same.

 

Edited January 21, 2023 by primerib1

[Wulfie Reanimator]

We don't have any kind of testing framework to work with, so test-driven development for LSL is impossible (or super impractical).

We can't trigger events (especially touch) or observe most behavior externally. The only way you could test anything is to have the tests in the same script as the implementation, which eats at your bytecode and active memory a lot.

Instead of having a listen event with LSL code in it, you would only have a bunch of user-function calls in the listen event. And for each user-function, you would also write multiple user-functions to test every aspect of that function.

 

6 hours ago, primerib1 said:

So that in the future as I add features / fix bugs / introduce optimizations, the externally-visible behavior of the script stays the same.

This is a very limited way of testing things and not worth the effort IMO.

Edited January 21, 2023 by Wulfie Reanimator

[v2nguard Dismantled]

Didn't there used to be a test environment for maybe Eclipse that could do basically all this? I feel like there was...

[Wulfie Reanimator]

1 hour ago, v2nguard Dismantled said:

Didn't there used to be a test environment for maybe Eclipse that could do basically all this? I feel like there was...

If you can find it, that'd be pretty cool.

While you could feasibly write a static analyzer for LSL, you can't really run tests on a script unless you could actually execute it. That would require someone to write their own LSL compiler/interpreter with extended functionality for triggering events. It seems very unlikely that anyone has or would do that.