[test] As dev, I'd like to know how to improve the test coverage of the package

[dagousket]

[test] As dev, I'd like to know how to improve the test coverage of the package so that all the critical steps of the model are thoroughly tested

Blocked

Some tests would require additional variables and datasets that are not included in the test data. Would it be possible to provide test data for the following objects ?

• to run fm_load_data() with time_step == "Quarter" , we would need the all_year variable
• to run fm_build_domain_mesh_spde() with create_mesh == "from_data" , we would need the study_domain_sf dataset
• to run fm_shape_sci_data_st() with scientific_observation == "Density" , we would need the Sci.obs_spp dataset

Decision

⚠ Given the number of untested parameters, we suggest that we implement as many as possible after solving the other problems raised during the recettage meeting.

• [ ] Please put "I accept" as a comment if you agree or detail what you think is essential for the future.

More information about what we have to test

By testing each parameter and keeping the other at test values :

✅ : this parameter is now tested

🟢 : ok to run with this parameter so we can add unit tests

🟠 : need rework on the function to test so you have to rework the code before testing

🔴 : need additional data so you have to send us data

• fm_load_data()
  • main_script
    • time_step == "Quarter" 🔴 missing all_year variable
    • fitted_data=="presabs" 🟠 needs to y_com_i as input
  • fm_build_domain_mesh_spde()
    • create_mesh == "from_data" 🔴 missing study_domain_sf dataset
  • fm_shape_sci_data_st()
    • scientific_observation == "Density" 🔴 missing Sci.obs_spp dataset
  • fm_shape_vms_logbook_data_st()
    • all covered ✅
• fm_fit_model()
  • main_script
    • T %in% str_detect(names(obj$par),"rho_" 🟠 obj$par is inner function
    • compute_sd ✅
  • fm_build_data_params_map()
    • "cov_fb"=if(is.null(xfb_x)){as.matrix(rep(1,nrow(cov_x_com)))}else{xfb_x} 🟢
    • "cov_fb_mesh"=if(is.null(xfb_x)){as.matrix(rep(1,mesh$n))}else{xfb_x} 🟢
    • "cov_fb_pred"=if(is.null(xfb_x)){as.matrix(rep(1,nrow(cov_x_pred)))}else{xfb_x} 🟢
    • data_source == 2 🟠 inner if condition on Version variable, set to "" by default
    • data_source == 3 ✅
    • biomass_temporal == 0 ✅
    • is.null(xfb_x) & samp_process==1 (covariate effects) 🟢
    • EM == "fix_b" (linkeage of density and sampling intensity) ✅
    • sampling_temporal == 1 (autoregressive process) ✅
    • T %in% str_detect(Random,"psiinput_x") ✅ (conditionned by sampling_temporal)
    • T %in% str_detect(Random,"par_b") ✅ (conditionned by EM)
    • samp_process == 1 ✅
    • data_source == 1 & ref_data == "sci" ✅
    • lf_link == 1 (precense-absence framework) ✅
• fm_generate_graphs()
  • main_script
    • nrow(time.step_df) > 1 ✅
    • samp_process == 1 ✅

Client - Validation

• [ ] The test coverage of the package has increased

• [ ] The missing test coverage detected in issue #66 and #70 are now included in the test

Dev - Tech

• [x] Pin the chunks of code without test coverage yet

• [x] report potential missing data/info to achieve test in these steps

• [ ] add test where needed

[balglave]

We provided modifications in the branch '91-blocked-client'. Let us know if this is fine.

[dagousket]

Hi @balglave , we have triggered a pull request for your branch to test your modifications on the CI. The CI fails to pass and hence prevents the merge of your edits. We suggest you check locally your version with devtools::check() to fix test failures prior to merging.