Antoine Fabri
Antoine Fabri
Currently we assume a default `initialise()` method so `new()` maps its inputs to attributes values, this is not general! ``` r track
We can't recycle args that have attributes that don't support recycling, which I think is all of those unless the class supports them. A use case that might have all...
facet_wrap works ok, not facet_grid, isn't this tested ? ``` r library(ggplot2) plt {constructive} couldn't create code that reproduces perfectly the input #> ℹ Call `construct_issues()` to inspect the last...
``` r x reticulate::r_to_py(1:3) x # x is actually an environment with a printing method #> [1, 2, 3] construct(py_to_r(x)) # but we can construct the equivalent R object, so...
After #527 * Add a method `.cstr_construct.python.builtin.object.as.environment()` so `construct(x, opts_environment("as.environment")` returns `as.environment(list(...))`. It's very similar to the first method implemented in #527, but instead of `py_to_r` we use `as.list` to...
After #527 At this point this should fail : `construct(structure(1, class = "python.builtin.object"))`, because our previous code assumes we construct "correct" objects. constructive has a mechanism to control that through...
After #528 `py_to_r()` and `r_to_py()` are not totally reliable, so #527 might break. I can reproduce it this way: ``` r library(reticulate) reticulate::use_virtualenv("r-reticulate") sentence_transformers
After #527 Because "python.builtin.object" objects contain pointers, `waldo::compare(x, y)` signals a difference even when `x` and `y` are equivalent. We can define methods for `waldo:: compare_proxy()` to edit the object...
This is similar to #527 "python.builtin.module" are also "python.builtin.objects" but they don't have R equivalents such as integer vectors or lists. We can however construct then with `import()`. No exported...
bytecodes can be created with `compiler::compile()` The code to use in `compile()` can be fetched from `res