RXR-1751 add safety

[roninsightrx]

When duplicate names are used in covariate and declare_variables arguments, the covariates are not exported in the output column. Example to see this:

library(PKPDsim)
library(pkvoriconazolefriberg)
mod <- pkvoriconazolefriberg::model()
covs <- list(
  WT = new_covariate(70),
  AGE = new_covariate(70),
  CYP2C19unknown = new_covariate(1),
  CYP2C19a1a2 = new_covariate(0),
  CYP2C19a2a2 = new_covariate(0),
  CYP2C19a2a3 = new_covariate(0),
  CYP2C19a1a3 = new_covariate(0),
  CYP2C19a3a3 = new_covariate(0)
)
reg <- new_regimen(amt = 500, interval = 12, n = 5)
res <- sim(
  mod,
  covariates = covs,
  regimen = reg,
  parameters = pkvoriconazolefriberg::parameters(),
  output_include = list(variables = TRUE, covariates = TRUE)
)
res[1:5,]

After this change it's now impossible to specify duplicate covariates and variables in model specs. I initially implemented code in sim() that checks for duplicates in the output of sim() and made the parsing safer, but I removed that because it should now be unnecessary.

• Added a test for throwing of the error
• removed PKPDsim.Rproj (unrelated cleanup)

[karawoo]

The code changes seems ok but I don't understand the reason behind removing the .Rproj file? Granted I don't use it since I don't use RStudio, but I thought most people do?

[roninsightrx]

The code changes seems ok but I don't understand the reason behind removing the .Rproj file? Granted I don't use it since I don't use RStudio, but I thought most people do?

We can discuss, but IMO .Rproj files shouldn't be included, I sere it as more of a user/environment specific file rather than something specific to the package. In this case it was annoying because it stores a flag that controls compilation (if all C++ source files should be recompiled. The current version has that flag set to true, but it takes 30 secs or so extra so annoying when developing). I could've set it to false but decided to remove instead.

[karawoo]

I think we should at least briefly discuss with the team to avoid breaking anyone's workflow for working with RStudio projects, because after this change RStudio's various project-specific options won't work (i.e. "open project" and "recent projects" won't work, you won't be able to double click on the .Rproj files to launch an RStudio session in that project, etc.).

Can we make it a separate PR from the RXR-1751 change?

[roninsightrx]

I think we should at least briefly discuss with the team to avoid breaking anyone's workflow for working with RStudio projects, because after this change RStudio's various project-specific options won't work (i.e. "open project" and "recent projects" won't work, you won't be able to double click on the .Rproj files to launch an RStudio session in that project, etc.).

Can we make it a separate PR from the RXR-1751 change?

Sounds good. I mean the idea is still that everyone (that uses RStudio) keeps the Rproj file, just not committed to git. We'll discuss tomorrow.

[roninsightrx]

this is the setting I mentioned. For PKPDsim it's better if this is set to no, since we don't want to keep compiling the C++ code