## Goal
Currently, there's no way in KCL to get fields of a plane, e.g. the underlying X axis, Y axis or origin.
This would be useful for geometry calculations in KCL. It would help KCL users write transformations between planes for rotating geometry.
For example, this enables
```kcl
export fn crossProduct(@vectors) {
a = vectors[0]
b = vectors[1]
x = a[1] * b[2] - (a[2] * b[1])
y = a[2] * b[0] - (a[0] * b[2])
z = a[0] * b[1] - (a[1] * b[0])
return [x, y, z]
}
export fn normalOf(@plane) {
return crossProduct([plane.xAxis, plane.yAxis])
}
```
## Implementation
My goal was just to enable a simple getter for planes, like `myPlane.xAxis` and yAxis and origins. That's nearly what happened, except I discovered that there's two ways to represent a plane: either `KclValue::Plane` or `KclValue::Object` with the right fields.
No matter which format your plane is represented as, it should behave consistently when you get its properties. Those properties should be returned as `[number; 3]` because that is how KCL represents points.
Unfortunately we actually require planes-as-objects to be defined with axes like `myPlane = { xAxis = { x = 1, y = 0, z = 0 }, ...}`, but that's a mistake in my opinion. So if you do use that representation of a plane, it should still return a [number; 3]. This required some futzing around so that we let you access object fields .x and .y as [0] and [1], which is weird, but whatever, I think it's good.
This PR is tested via planestuff.kcl which has a Rust unit test.
Part of the hole efforts, see https://github.com/KittyCAD/modeling-app/discussions/7543
* Change to use deterministic artifact graph
* Update output to use the new order
* Fix to clear everything when scene is cleared
* Fix lots
* Update artifact graph output for the last time
* Delete unused sorting code
* Remove unneeded cfg
* Fix to preserve top-level artifacts when there's an error
* Update output after error fix
* Add better doc comments
* Remove duplicate global operations
* Update comments
* Update ignored tests that were flaky
* Update graph for new samples after rebase
* Fix test assertion message
* Improve docs around PI
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Refactor and polish type error messages
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Add suggestion to fix unknown numbers error
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Don't warn so often about unknown units
Signed-off-by: Nick Cameron <nrc@ncameron.org>
---------
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Add degrees annotations to examples
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Fix a units bug with the modulo operation
Signed-off-by: Nick Cameron <nrc@ncameron.org>
---------
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Change so that operations are stored per module
* Refactor so that all modeling commands go through ExecState
* Remove unneeded PartialOrd implementations
* Remove artifact_commands from KclError since it was only for debugging
---------
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Previously in a member expression like `foo.x` or `foo[3]`, `foo` had to be an identifier. You could not do something like `f().x` (and if you tried, you got a cryptic error). Rather than make the error better, we should just accept any expression to be the LHS of a member expression (aka its 'object').
This does knock our "parse lots of function calls" from 58 to 55 calls before it stack overflows. But I think it's fine, we'll address this in https://github.com/KittyCAD/modeling-app/pull/6226 when I get back to it.
Closes https://github.com/KittyCAD/modeling-app/issues/7273
I ignored some new clippy lints about large differences between enum variants.
We can always revisit these later (the compiler suggests boxing them so
that the enum variants are similar size)
Previously, `x = cos(x)` would just say "`x` is undefined". Now it says that `x` cannot be referenced in its own definition, try using a different variable instead.
To do this, I've added a new `Option<String>` field to the mod-local executor context, tracking the current variable declaration. This means cloning some strings, implying a small performance hit. I think it's fine, for the better diagnostics.
In the future we could refactor this to use a &str or store variable labels in stack-allocated strings like docs.rs/compact_str or something.
Closes https://github.com/KittyCAD/modeling-app/issues/6072
We've changed the unnamed field of `KclError` variants to a named called `details`.
To clarify: previously KCL errors looked like this:
```rust
pub enum KclError {
Lexical(KclErrorDetails),
Syntax(KclErrorDetails),
```
Now they look like this:
```rust
pub enum KclError {
Lexical { details: KclErrorDetails },
Syntax { details: KclErrorDetails },
}
```
This lets us more easily add fields to the errors. For example, in the UndefinedValue case, adding a field for what the undefined name was. This PR refactors the code to make my PR in https://github.com/KittyCAD/modeling-app/pull/7309 much easier.
Pure refactor, should not change any behaviour.
* Add backtrace to errors
* Add display of backtraces with hints
* Change pane badge to only show count of errors
* Fix property name to not collide with Error superclass
* Increase min stack again
* Add e2e test that checks that the diagnostics are created in CodeMirror
* Remove unneeded code
* Change to the new hotness
* Shuffle around function call code
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Refactor function calls to share more code
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Hack to leave the result of revolve as a singleton rather than array
Signed-off-by: Nick Cameron <nrc@ncameron.org>
---------
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Treat number as any rather than default
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Don't square root negative numbers
Signed-off-by: Nick Cameron <nrc@ncameron.org>
---------
Signed-off-by: Nick Cameron <nrc@ncameron.org>
* Add sim test for any type
* Fix doc comments to match code
* Add array ascription tests
* Commit new test output
* Fix to not panic when type is undefined
* Fix to not panic on use of the any type
* Update test and generated output
* Fix error message after rebase
* Fix subtype of any
* Fix KCL to use new keyword args
* Fix to not nest MixedArray in HomArray
* Update output
* Remove all creation of MixedArray and use HomArray instead
* Rename MixedArray to Tuple
* Fix to coerce arrays the way tuples are done
* Restructure to appease the type signature extraction
* Fix TS unit test
* Update output after switch to HomArray
* Update docs
* Fix to remove edge case when creating points
* Update docs with broken point signature
* Fix display of tuples to not collide with arrays
* Change push to an array with type mismatch to be an error
* Add sim test for push type error
* Fix acription to more general array element type
* Fix to coerce point types
* Change array push to not error when item type differs
* Fix coercion tests
* Change to only flatten as a last resort and remove flattening tuples
* Contort code to appease doc generation
* Update docs
* Fix coerce axes
* Fix flattening test to test arrays instead of tuples
* Remove special subtype case for singleton coercion
* Change array functions to call user function with keyword args
* Fix KCL to use keyword params
* Remove unneeded positional call code
* Update docs
* Update output
Preparing for the removal of positional functions from the language. The first big step is to change all our KCL code examples, test code, public samples etc to all use keyword functions.
Apologies for how large this PR is. Most of it is:
- Changing example KCL that defined its own functions, so the functions now use keyword arguments rather than positional arguments. E.g. change `cube([20, 20])` to be `cube(center = [20, 20])`.
- Some parts of the code assumed positional code and didn't handle keyword calls, e.g. the linter would only check for positional calls to startSketchOn. Now they should work with either positional or keyword.
- Update all the artifacts
This does _not_ remove support for positional calls. That will be in a follow-up PR.
* Change so that var definitions can be a module's return value
* Change car wheel assembly to use the new return mechanism
* Add sim test
* Update output
* Update module docs
* Add safety check to only work with modules
* Fix to use updated keyword args
