Create a function library for the evaluator, bring back the ICCC voltage divider#16
Conversation
luxas
left a comment
luxas
left a comment
There was a problem hiding this comment.
Nicely done! The resistor-calc crate is great 😄
| @@ -0,0 +1,11 @@ | |||
| [package] | |||
| name = "kicad_functions" | |||
There was a problem hiding this comment.
Any specific reason to make this a new crate compared to build into kicad_rs?
There was a problem hiding this comment.
Mostly just for easy expandability, one does not need to touch core application logic to extend its functionality (via adding these functions). IIRC the "external library" approach was also preferred when I asked about it.
| F 2 "Capacitor_SMD:C_1206_3216Metric" H 3500 4000 50 0001 C CNN | ||
| F 3 "~" H 3500 4000 50 0001 C CNN | ||
| F 4 "" H 3500 4000 50 0001 C CNN "DividerTest" | ||
| F 5 "vdiv(5.1, \"(R1+R2)/R2*0.8\", \"E96\", 500e3, 700e3)" H 3500 4000 50 0001 C CNN "DividerTest_expr" |
There was a problem hiding this comment.
I'm a bit unclear on how to use this correctly, is it a map you index like vdiv(...)[2] to get the R2 voltage?
There was a problem hiding this comment.
The function returns a tuple, it's described in the docs of the voltage_divider function. You need to store the result to a variable (like DividerTest here, but preferrably stored globally) to avoid running the (potentially very long) computation over and over again. What I have indeed not tested is indexing the evalexpr::Value(TupleType), but I sure hope that evalexpr supports indexing its own tuples.
| if let Some(res) = calculate(&config) { | ||
| // Take the first result, these are ordered by increasing error | ||
| if let Some((v, set)) = res.iter().next() { | ||
| let voltage = config.target + *v as f64 / 1e9; |
There was a problem hiding this comment.
Taken straight from resistor-calc, it computes using fixed-point numbers (u64) that need to be converted to floats explicitly.
There was a problem hiding this comment.
Can you add parentheses to signal in what order the computation happens and a comment why this is like this (e.g. a link to another place this constant is used)? Or better, make this a dedicated helper
There was a problem hiding this comment.
I've added a dedicated helper with an explanation now.
| fn parse(v: &Value) -> EvalexprResult<Self> { | ||
| let tuple = v.as_tuple()?; | ||
|
|
||
| let resistance_min = tuple.get(3).map(|v| v.as_number()).transpose()?; |
| } | ||
| } | ||
|
|
||
| fn unescape(s: &str) -> String { |
There was a problem hiding this comment.
Is this a "custom" unescaping function, or is it a "conventional" like done for e.g. URL escaping?
There was a problem hiding this comment.
Its custom, I just cobbled it together since kicad_parse_gen messes up escaping so badly.
| for i in 1..=config.count { | ||
| tuple.push(Value::from(set.r(i))); | ||
| } | ||
| return Ok(Value::from(tuple)); |
There was a problem hiding this comment.
Does evalexpr support tuples natively, or are they arrays?
There was a problem hiding this comment.
The "tuple" is just a Vec with any amount of elements. evalexpr has a an enum variant Value(TupleType), where TupleType is an alias for Vec<Value>. I haven't checked what is supported for those though, i.e. indexing is pretty much required for this use case. Will check in a bit if that's implemented.
|
After a bit of investigation it doesn't seem like there's built-in support for indexing tuples. I'll create an indexing function, hold on... |
twelho
force-pushed
the
evaluator-functions
branch
from
9a9a43f to
909923a
Compare
twelho
force-pushed
the
evaluator-functions
branch
2 times, most recently
from
c65bd24 to
aeb3f3a
Compare
2 participants
This moves
kicad_rsand the new function librarykicad_functionsinto a common Cargo workspace to facilitate dependency version sharing as well as make running stuff likecargo fmtfor the project easier.The new voltage divider function is based on https://github.com/trebinor/resistor-calc, which I had to fork and modify a bit to get to compile without its own expression evaluator. PR tracking upstreaming here: trebinor/resistor-calc#4.
Support list for
voltage_divider:Continuation of #14, ticking one more box in the support list there. The last commit 9a9a43f is just test data for evaluating the voltage divider and may be removed before merging. That test data configuration is identical to what was passed to ICCC for the BD9E DC-DC converter, and this Rust code does output the same optimal resistance values, so it should be working correctly.
cc @luxas @chiplet