Brainstorm ways to automatically update save_all_plots() #432

Activity

1. Schiano-NOAA

   added this to {stockplotr} FY27 Goals and Timelineon Sep 29, 2026
2. JZahner1

   assigned

   Schiano-NOAA

   and

   sbreitbart-NOAA

   on Oct 20, 2026
3. JZahner1

   converted this from a draft issue on Oct 20, 2026
4. JZahner1 commented on Oct 20, 2026

   

   JZahner1

   on Oct 20, 2026

   CollaboratorAuthor

   More actions

   So, I think there are, nominally, two ways to think about doing this.

   The first is basically to make the function take, as input parameters, lists that correspond to each plot/table function and contain function specific parameters. For example:

   save_all_plots(
       dat,
       plot_recruitment_options =list(
           module = "TIME_SERIES",
           unit_label = "mt", 
           scale_amount = 1
       ),
       plot_biomass_options = list(
           relative = FALSE,
           ref_line = "msy",
           unit_label = "mt",
           scale_amount = 1
       ),
       ...,
       interactive=FALSE,
       figures_tables_dir=getwd()
   )

   Each plot/table specific list could then contain elements corresponding to the input parameters for that function. This is readily extensible and would also allow users to customize plot specific module parameters easily. This does have the downside of requiring adding new default parameters when adding a function, but I dont know of a simple way around that. I have tested this set up and it works, just maybe requires a lot from the user and would probably get unwieldy with more plots.

   The second way to do this would involve doing something like below:

   save_all_plots(
       dat,
       module = c("plot_recruitment"="TIME_SERIES"),
       unit_label = c("plot_recruitment"="mt", "plot_biomass"="mt", ...)
       scale_amount = c("plot_recruitment"=1, "plot_biomass"=1, ...)
       ...,
       interactive=FALSE,
       figures_tables_dir=getwd()
   )

   Here, we would maintain the standard plotting interface of the individual functions, but users would specify plot specific customizations using a named vector. I have not fully tested this style yet, but dont see why it would not conceptually work.

   Regardless of how the inputs are set up, the function itself becomes a pretty simple loop over all available plot_* and table_* functions, with some care taken to set the function specific parameter arguments depending on the input style. It would looke something like:

   for(i in seq_along(plottab_functions)){
         func_name <- plottab_functions[i]
         # print(all_params)
         func_specific_args <- all_params[paste0(func_name,"_options")][[1]]
         func_pars <- c(default_args, func_specific_args)
   
         if(grepl("table",func_name)){
           names(func_pars)[which(names(func_pars)=="figures_dir")] <- "tables_dir"
         }
         # print(func_pars)
   
         cli::cli_alert_info("Attempting to execute function: {func_name}")
         tryCatch(
           {
             suppressMessages(
                 do.call(func_name, func_pars) # Execute plotting/table function
             )
           },
           error = function(e) {
               cli::cli_alert_danger("{func_name} failed to run.")
               cli::cli_alert("Tip: check that your arguments are correct.")
               for(a in func_specific_args){
                 arg_name = names(func_specific_args)[a]
                 arg_value = as.character(func_specific_args[a])
                 cli::cli_li("{arg_name} = {arg_value}")
               }
               print(e)
           }
         )   
     }

   Thoughts or strong opinions on either of these strategies?
5. Schiano-NOAA commented on Oct 20, 2026

   

   Schiano-NOAA

   on Oct 20, 2026

   Last edited by Schiano-NOAA

   Collaborator

   More actions

   I think I am leaning towards the first more. For this issue, I was more thinking along the lines of trying to automate adding a plot to save_all_plots(). Like creating a GitHub action that would append the new function and format into save_all_plots then open a PR. It would then serve as (1) a reminder to add the plot or figure to the function and (2) a way to add it in without putting much thought. The reviewer of the PR would then need to just adjust input values if it varies from the default.

   I think what you said could still be used and applied to the GHA, but it's more of a revamp of the current function if that makes sense. Either way, I like the proposed approaches 😄
6. JZahner1 commented on Oct 20, 2026

   

   JZahner1

   on Oct 20, 2026

   CollaboratorAuthor

   More actions

   @Schiano-NOAA oh okay. Yah, it definitely could be a github action if that's more what you intended. Im not familiar with enough with how GitHub actions works to make that sort of change.

   I do wonder whether this type of user input is still necessary for the actual R function. I can imagine scenario where users might want to set a few custom things for a few plots but otherwise get the default versions of everything else.
7. JZahner1 commented on Oct 21, 2026

   

   JZahner1

   on Oct 21, 2026

   CollaboratorAuthor

   More actions

   I should also note that the first implementation above relies on there only being a single function per file, and that the function name and file name match exactly. For example, I do not believe it would work alongside the composition plots in #424, as is.
8. github-project-automation

   moved this from Todo to In progress in {stockplotr} FY27 Goals and Timelineon Oct 22, 2026