expression as actual expression (with environment) or quosure?

[joethorley]

• Need to patch https://github.com/poissonconsulting/mbr, e.g., mbr:::mcmc_derive.mb_analysis() and mbr:::mcmc_derive_fun(): use {{ new_expr }} instead of new_expr

[joethorley]

This was reraised during MCMC athon and is now a priority.

[joethorley]

Need to retain ability to work with old strings.

[krlmlr]

Reprex to see which code is seeing new_expr :

options(conflicts.policy = list(warn = FALSE))
devtools::load_all(".")
#> ℹ Loading mcmcrtesting

sbf_set_sub("count")
analysis <- sbf_load_object("analysis")
#> Registered S3 method overwritten by 'rjags':
#>   method               from 
#>   as.mcmc.list.mcarray mcmcr
sbf_load_datas()

# slow part appears to be under within.list | eval | res_gamma_pois
# cbind/vectorized with residual ----
# this is fast and getting to where we want

mcmc_derive_data(
  analysis,
  new_data = data,
  new_expr = "
    rlang::abort('Just for testing')
    log(eCount) <- b0 + bYear * Year + bKelpLine * KelpLine + bSite[Site] + bAnnual[Annual] + bSiteAnnual[cbind(Site, Annual)]
    log(eCountKelpline) <- b0 + bKelpLine + bYear * Year + bAnnual[Annual] + bSite[Site] + bSiteAnnual[cbind(Site,Annual)]
    log(eCountBarren) <- b0 + bYear * Year + bAnnual[Annual] + bSite[Site] + bSiteAnnual[cbind(Site,Annual)]
    prediction <- eCountKelpline * 0.25 + eCountBarren * 0.75

    residual <- res_gamma_pois(Count, fit, sSiteAnnualQuadrat)
"
)
#> Error:
#> ! Just for testing
#> Backtrace:
#>      ▆
#>   1. ├─mcmcdata::mcmc_derive_data(analysis, new_data = data, new_expr = "\n    rlang::abort('Just for testing')\n    log(eCount) <- b0 + bYear * Year + bKelpLine * KelpLine + bSite[Site] + bAnnual[Annual] + bSiteAnnual[cbind(Site, Annual)]\n    log(eCountKelpline) <- b0 + bKelpLine + bYear * Year + bAnnual[Annual] + bSite[Site] + bSiteAnnual[cbind(Site,Annual)]\n    log(eCountBarren) <- b0 + bYear * Year + bAnnual[Annual] + bSite[Site] + bSiteAnnual[cbind(Site,Annual)]\n    prediction <- eCountKelpline * 0.25 + eCountBarren * 0.75\n\n    residual <- res_gamma_pois(Count, fit, sSiteAnnualQuadrat)\n")
#>   2. └─mbr:::mcmc_derive_data.mb_analysis(...)
#>   3.   ├─mcmcderive::mcmc_derive(...)
#>   4.   └─mbr:::mcmc_derive.mb_analysis(...) at mcmcderive/R/mcmc-derive.R:34:2
#>   5.     └─mbr:::mcmc_derive_fun(...)
#>   6.       ├─mcmcderive::mcmc_derive(...)
#>   7.       └─mcmcderive:::mcmc_derive.mcmcr(...) at mcmcderive/R/mcmc-derive.R:34:2
#>   8.         └─mcmcderive:::split_apply_combine(...) at mcmcderive/R/mcmc-derive.R:117:2
#>   9.           └─base::lapply(...) at mcmcderive/R/internal.R:167:4
#>  10.             └─mcmcderive (local) FUN(X[[i]], ...)
#>  11.               └─base::lapply(...) at mcmcderive/R/internal.R:146:2
#>  12.                 └─mcmcderive (local) FUN(X[[i]], ...)
#>  13.                   ├─base::within(object, eval(expr)) at mcmcderive/R/internal.R:137:2
#>  14.                   └─base::within.list(object, eval(expr))
#>  15.                     └─base::eval(substitute(expr), e)
#>  16.                       └─base::eval(substitute(expr), e)
#>  17.                         ├─base::eval(expr)
#>  18.                         │ └─base::eval(expr)
#>  19.                         └─rlang::abort("Just for testing")

^{Created on 2023-05-25 with reprex v2.0.2}