Remove get_stats_path
#230
Labels
enhancement
New feature or request
Comments
bors bot
added a commit
that referenced
this issue
Feb 28, 2022
231: Prep for removing get_stats_path r=charleskawczynski a=charleskawczynski A step towards #230. Co-authored-by: Charles Kawczynski <[email protected]>
4 tasks
bors bot
added a commit
that referenced
this issue
May 3, 2022
331: Store z_obs vector and reference namelist in ReferenceModel. r=ilopezgp a=ilopezgp ## Changes Constructs the vector of observations and reference namelist at ReferenceModel construction time and stores them within the struct. This approach has a number of advantages: - Reduces number of calls to the infamous `get_stats_path`, - Stores observational grid points within ReferenceModel, which enables writing them to the Diagnostics. Right now it was only possible to reconstruct them implicitly based on constant `dz` assumptions, which we want to move away from, - Enables passing non-default grid namelist arguments separately to each ReferenceModel. This allows greater flexibility than the current approach, in which we either calibrated to the default resolution or to a common resolution for all cases. - Removes the need to use `run_reference_SCM`! We retain the function to be able to call TC.jl without calibration, which may be useful. But `run_reference_SCM` is no longer used in the calibration pipeline. - Reduces the time spent in the initialization step. ## Issue number (if applicable) This closes issue #319, and should be a helpful step towards resolving #311 and #230. ## Checklist before requesting a review / merging - [x] I have formatted the code using `julia --project=.dev .dev/climaformat.jl .` - [x] I have updated all test manifests using `julia --project .dev/up_deps.jl .` with Julia 1.7.0. - [x] If core features are added, I have added appropriate test coverage. - [x] If new functions and structs are added, they are documented through docstrings. Co-authored-by: Ignacio Lopez Gomez <[email protected]>
bors bot
added a commit
that referenced
this issue
May 3, 2022
331: Store z_obs vector and reference namelist in ReferenceModel. r=ilopezgp a=ilopezgp ## Changes Constructs the vector of observations and reference namelist at ReferenceModel construction time and stores them within the struct. This approach has a number of advantages: - Reduces number of calls to the infamous `get_stats_path`, - Stores observational grid points within ReferenceModel, which enables writing them to the Diagnostics. Right now it was only possible to reconstruct them implicitly based on constant `dz` assumptions, which we want to move away from, - Enables passing non-default grid namelist arguments separately to each ReferenceModel. This allows greater flexibility than the current approach, in which we either calibrated to the default resolution or to a common resolution for all cases. - Removes the need to use `run_reference_SCM`! We retain the function to be able to call TC.jl without calibration, which may be useful. But `run_reference_SCM` is no longer used in the calibration pipeline. - Reduces the time spent in the initialization step. ## Issue number (if applicable) This closes issue #319, and should be a helpful step towards resolving #311 and #230. ## Checklist before requesting a review / merging - [x] I have formatted the code using `julia --project=.dev .dev/climaformat.jl .` - [x] I have updated all test manifests using `julia --project .dev/up_deps.jl .` with Julia 1.7.0. - [x] If core features are added, I have added appropriate test coverage. - [x] If new functions and structs are added, they are documented through docstrings. Co-authored-by: Ignacio Lopez Gomez <[email protected]>
bors bot
added a commit
that referenced
this issue
May 3, 2022
331: Store z_obs vector and reference namelist in ReferenceModel. r=ilopezgp a=ilopezgp ## Changes Constructs the vector of observations and reference namelist at ReferenceModel construction time and stores them within the struct. This approach has a number of advantages: - Reduces number of calls to the infamous `get_stats_path`, - Stores observational grid points within ReferenceModel, which enables writing them to the Diagnostics. Right now it was only possible to reconstruct them implicitly based on constant `dz` assumptions, which we want to move away from, - Enables passing non-default grid namelist arguments separately to each ReferenceModel. This allows greater flexibility than the current approach, in which we either calibrated to the default resolution or to a common resolution for all cases. - Removes the need to use `run_reference_SCM`! We retain the function to be able to call TC.jl without calibration, which may be useful. But `run_reference_SCM` is no longer used in the calibration pipeline. - Reduces the time spent in the initialization step. ## Issue number (if applicable) This closes issue #319, and should be a helpful step towards resolving #311 and #230. ## Checklist before requesting a review / merging - [x] I have formatted the code using `julia --project=.dev .dev/climaformat.jl .` - [x] I have updated all test manifests using `julia --project .dev/up_deps.jl .` with Julia 1.7.0. - [x] If core features are added, I have added appropriate test coverage. - [x] If new functions and structs are added, they are documented through docstrings. Co-authored-by: Ignacio Lopez Gomez <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
get_stats_pathis a very complicated function for something that should probably be handled by callers in a much simpler way. It seems that, in most cases,get_stats_pathshould simply returnjoinpath(dir, "stats", "$case_name.nc"), and the existence of this file, which is checked in this function, could be checked much earlier.The text was updated successfully, but these errors were encountered: