transition update internals
#704
Conversation
no need to snapshot the `print.parameters` method: it may change soon, and some extension packages aren't using 3e yet
this way, we can `Remotes` a specific parsnip version before package releases
|
re: follow-up discussion from the group meeting, going to go ahead and work some A la Line 103 in 68dfd60 |
The process:
* put together a `update_spec` helper that does most all of the work that the `update` methods do
* for each update method:
+ drop in the `update_spec` template
+ update it’s `cls` argument to match the same `new_model_spec` argument
+ if the `update` method doesn’t take a parameters argument, set it to NULL in `update_spec`
+ delete remaining code besides the `args` `enquo`ting lines
Push before updating any of the tests to make sure that checks pass on the previous tests too.
`fresh` was ignored and engine arguments propagated through `update`
transition most tests to `test_update` and test more thoroughly with one example. for each update method test, pass both a model and engine argument, and then `update` both of them to `tune()` (with some exceptions depending on available arguments).
|
Okay, good to go! This PR now refactors the internals of One thing worth mentioning—I'm not sure if this was intended behavior, but pak::pkg_install("tidymodels/parsnip")
#> ℹ Loading metadata database
#> ℹ Loading metadata database
library(parsnip)
linear_reg(mixture = 0) %>%
set_engine("glmnet", nlambda = 10) %>%
update(penalty = 1, fresh = TRUE)
#> Linear Regression Model Specification (regression)
#>
#> Main Arguments:
#> penalty = 1
#>
#> Engine-Specific Arguments:
#> nlambda = 10
#>
#> Computational engine: glmnetCreated on 2022-04-21 by the reprex package (v2.0.1) It now handles engine args in the same way we've previously handled main ones: pak::pkg_install("tidymodels/parsnip@engine-params-691")
#> ℹ Loading metadata database
#> ℹ Loading metadata database
library(parsnip)
linear_reg(mixture = 0) %>%
set_engine("glmnet", nlambda = 10) %>%
update(penalty = 1, fresh = TRUE)
#> Linear Regression Model Specification (regression)
#>
#> Main Arguments:
#> penalty = 1
#>
#> Computational engine: glmnetCreated on 2022-04-21 by the reprex package (v2.0.1) 🐝 |
That was not intended. Can you update this PR with that change? |
|
@topepo That change is already made here! Added a bullet to NEWS re: that fix—give me a holler if that's not what you meant. |
|
Glad to fix merge conflicts if this PR is otherwise good to go!🐨 |
|
This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue. |
See #691 and #691 (comment). :)