don't write rdata in test mode, improve start scripts

CHANGELOG.md

@@ -28,6 +28,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
     [[#1414](https://github.com/remindmodel/remind/pull/1414)]
 - correctly report `Tech|*|Capital Costs|w/ Adj Costs` for t < cm_startyear
     [[#1429](https://github.com/remindmodel/remind/pull/1429), [#1476](https://github.com/remindmodel/remind/pull/1476)]
+- **scripts** '--test' mode for start.R and start_bundle_coupled.R does not write RData files anymore
+    [[#1500](https://github.com/remindmodel/remind/pull/1500)]
 - prevent tradtional biomass spillover to other sectors than buildings
     [[#1519](https://github.com/remindmodel/remind/pull/1519)]
 

scripts/start/readCheckScenarioConfig.R

@@ -102,7 +102,7 @@ readCheckScenarioConfig <- function(filename, remindPath = ".", testmode = FALSE
     BAUbutNotNeeded <- ! is.na(scenConf$path_gdx_bau) & ! (scenNeedsBau)
     if (sum(BAUbutNotNeeded) > 0) {
       msg <- paste0("In ", sum(BAUbutNotNeeded), " scenarios, 'path_gdx_bau' is not empty although no realization is selected that needs it.\n",
-                    "To avoid unnecessary dependencies to other runs, setting 'path_gdx_bau' to NA for:\n",
+                    "To avoid unnecessary dependencies to other runs, automatically setting 'path_gdx_bau' to NA for:\n",
                     paste(rownames(scenConf)[BAUbutNotNeeded], collapse = ", "))
       message(msg)
       scenConf$path_gdx_bau[BAUbutNotNeeded] <- NA
@@ -132,6 +132,7 @@ readCheckScenarioConfig <- function(filename, remindPath = ".", testmode = FALSE
   }
   unknownColumnNames <- names(scenConf)[! names(scenConf) %in% knownColumnNames]
   if (length(unknownColumnNames) > 0) {
+    message("")
     forbiddenColumnNames <- list(   # specify forbidden column name and what should be done with it
        "c_budgetCO2" = "Rename to c_budgetCO2from2020, adapt emission budgets, see https://github.com/remindmodel/remind/pull/640",
        "c_budgetCO2FFI" = "Rename to c_budgetCO2from2020FFI, adapt emission budgets, see https://github.com/remindmodel/remind/pull/640",

start.R

@@ -34,8 +34,7 @@ helpText <- "
 #'                       the full model.
 #'   --reprepare, -R:    rewrite full.gms and restart run
 #'   --restart, -r:      interactively restart run(s)
-#'   --test, -t:         test scenario configuration and writing the RData files
-#'                       in the REMIND main folder without starting the runs
+#'   --test, -t:         test scenario configuration without starting the runs
 #'   --testOneRegi, -1:  starting the REMIND run(s) in testOneRegi mode
 #'   startgroup=MYGROUP  when reading a scenario config .csv file, don't start
 #'                       everything specified by \"start = 1\", instead start everything
@@ -74,15 +73,13 @@ if(!exists("argv")) argv <- commandArgs(trailingOnly = TRUE)
 argv <- argv[! grepl("^-", argv) & ! grepl("=", argv)]
 # check if user provided any unknown arguments or config files that do not exist
 if (length(argv) == 1) {
-  if (file.exists(argv)) {
-    config.file <- argv
-  } else if (file.exists(file.path("config", argv))) {
-    config.file <- file.path("config", argv)
-  } else {
-    stop("Unknown parameter provided: ", paste(argv, collapse = ", "))
-  }
+  config.file <- argv
+  if (! file.exists(config.file)) config.file <- file.path("config", argv)
+  if (! file.exists(config.file)) config.file <- file.path("config", paste0("scenario_config_", argv, ".csv"))
+  if (! file.exists(config.file)) stop("Unknown parameter provided: ", paste(argv, collapse = ", "))
 } else if (length(argv) > 1) {
-  stop("You provided more than one file or other command line argument, start.R can only handle one.")
+  stop("You provided more than one file or other command line argument, start.R can only handle one: ",
+       paste(argv, collapse = ", "))
 }
 
 if ("--help" %in% flags) {
@@ -333,7 +330,7 @@ if (any(c("--reprepare", "--restart") %in% flags)) {
     }
 
     # save the cfg object for the later automatic start of subsequent runs (after preceding run finished)
-    if (! "--gamscompile" %in% flags) {
+    if (! any(c("--test", "--gamscompile") %in% flags)) {
       filename <- paste0(cfg$title,".RData")
       message("   Writing cfg to file ", filename)
       save(cfg, file=filename)
@@ -366,12 +363,14 @@ if (any(c("--reprepare", "--restart") %in% flags)) {
   if (exists("lockID")) gms::model_unlock(lockID)
 }
 
+warnings()
+
 message("\nFinished: ", startedRuns, " runs started. ", waitingRuns, " runs are waiting. ",
         if (modeltestRunsUsed > 0) paste0(modeltestRunsUsed, " GDX files from modeltests selected."))
 if ("--gamscompile" %in% flags) {
   message("To investigate potential FAILs, run: less -j 4 --pattern='^\\*\\*\\*\\*' filename.lst")
 } else if ("--test" %in% flags) {
-  message("You are in --test mode: Rdata files were written, but no runs were started. ", errorsfound, " errors were identified.")
+  message("You are in --test mode: no runs were started. ", errorsfound, " errors were identified.")
 } else if (model_was_locked & (! "--restart" %in% flags | "--reprepare" %in% flags)) {
   message("The model was locked before runs were started, so they will have to queue.")
 }

start_bundle_coupled.R

@@ -32,18 +32,17 @@ helpText <- "
 #'
 #' Control the script's behavior by providing additional arguments:
 #'
-#'   --help, -h:        show this help text and exit
-#'   --gamscompile, -g: compile gms of all selected runs. Combined with
-#'                      --interactive, it stops in case of compilation
-#'                      errors, allowing to fix them and rerun gamscompile
-#'   --interactive, -i: interactively select run(s) to be started. Asks for
-#'                      config file also if the one specified as
-#'                      path_settings_coupled cannot be found.
-#'   --test, -t:        Test scenario configuration and write the RData files
-#'                      in the REMIND main folder without starting the runs.
+#'   --help, -h:         show this help text and exit
+#'   --gamscompile, -g:  compile gms of all selected runs. Combined with
+#'                       --interactive, it stops in case of compilation
+#'                       errors, allowing to fix them and rerun gamscompile
+#'   --interactive, -i:  interactively select run(s) to be started. Asks for
+#'                       config file also if the one specified as
+#'                       path_settings_coupled cannot be found.
+#'   --test, -t:         Test scenario configuration without starting the runs.
 #'   startgroup=MYGROUP  when reading a scenario config .csv file, don't start
-#'                       everything specified by \"start = 1\", instead start everything
-#'                       specified by \"start = MYGROUP\"
+#'                       everything specified by \"start = 1\", instead start
+#'                       everything specified by \"start = MYGROUP\"
 "
 
 ########################################################################################################
@@ -523,7 +522,7 @@ for(scen in common){
       # runs by hand after the NDC has finished.
     }
     if (i == start_iter_first && ! start_now && all(file.exists(cfg_rem$files2export$start[path_gdx_list]) | unlist(gdx_na))) {
-        start_now <- TRUE
+      start_now <- TRUE
     }
     foldername <- file.path("output", fullrunname)
     if ((i > start_iter_first || !scenarios_coupled[scen, "start_magpie"]) && file.exists(foldername)) {
@@ -551,11 +550,15 @@ for(scen in common){
       errorsfound <- errorsfound + cfg_rem$errorsfoundInCheckFixCfg
     }
 
-    Rdatafile <- paste0(fullrunname, ".RData")
-    message("Save settings to ", Rdatafile)
-    save(path_remind, path_magpie, cfg_rem, cfg_mag, runname, fullrunname, max_iterations, start_iter,
-         n600_iterations, path_report, qos, sbatch, prefix_runname, run_compareScenarios, magpie_empty,
-         numberOfTasks, start_now, file = Rdatafile)
+    if (! "--test" %in% flags) {
+      Rdatafile <- paste0(fullrunname, ".RData")
+      message("Save settings to ", Rdatafile)
+      save(path_remind, path_magpie, cfg_rem, cfg_mag, runname, fullrunname, max_iterations, start_iter,
+           n600_iterations, path_report, qos, sbatch, prefix_runname, run_compareScenarios, magpie_empty,
+           numberOfTasks, start_now, file = Rdatafile)
+    } else if (start_now) {
+      startedRuns <- c(startedRuns, fullrunname)
+    }
 
   } # end for (i %in% iterations)
 
@@ -604,12 +607,16 @@ for(scen in common){
 # start runs
 message("\nStarting Runs")
 for (scen in common) {
-  if (!scenarios_coupled[scen, "start_scenario"]) {
+  if (! scenarios_coupled[scen, "start_scenario"]) {
     next
   }
   start_iter_first <- scenarios_coupled[scen, "start_iter_first"]
   runname <- paste0(prefix_runname, scen)
   fullrunname <- paste0(runname, "-rem-", start_iter_first)
+  if ("--test" %in% flags || "--gamscompile" %in% flags) {
+    message("Test mode: run ", fullrunname, " NOT submitted to the cluster.")
+    next
+  }
   Rdatafile <- paste0(fullrunname, ".RData")
   runEnv <- new.env()
   load(Rdatafile, envir = runEnv)
@@ -623,26 +630,22 @@ for (scen in common) {
         runEnv$qos <- gsub("^(medium|standby)$", "auto", runEnv$qos)
         runEnv$numberOfTasks <- 3
       }
-      if ("--test" %in% flags || "--gamscompile" %in% flags) {
-        message("Test mode: run ", fullrunname, " NOT submitted to the cluster.")
-      } else {
-        logfile <- file.path("output", fullrunname, paste0("log", if (scenarios_coupled[scen, "start_magpie"]) "-mag", ".txt"))
-        if (! file.exists(dirname(logfile))) dir.create(dirname(logfile))
-        message("Find logging in ", logfile)
-        if (isTRUE(runEnv$qos == "auto")) {
-          sq <- system(paste0("squeue -u ", Sys.info()[["user"]], " -o '%q %j'"), intern = TRUE)
-          runEnv$qos <- if (is.null(attr(sq, "status")) && sum(grepl("^priority ", sq)) < 4) "priority" else "short"
-        }
-        slurmOptions <- combine_slurmConfig(paste0("--qos=", runEnv$qos, " --job-name=", fullrunname, " --output=", logfile,
-          " --open-mode=append --mail-type=END --comment=REMIND-MAgPIE --tasks-per-node=", runEnv$numberOfTasks,
-          if (runEnv$numberOfTasks == 1) " --mem=8000"), runEnv$sbatch)
-        slurmCommand <- paste0("sbatch ", slurmOptions, " --wrap=\"Rscript start_coupled.R coupled_config=", Rdatafile, "\"")
-        message(slurmCommand)
-        exitCode <- system(slurmCommand)
-        if (0 < exitCode) {
-          errorsfound <- errorsfound + 1
-          message("sbatch command failed, check logs.")
-        }
+      logfile <- file.path("output", fullrunname, paste0("log", if (scenarios_coupled[scen, "start_magpie"]) "-mag", ".txt"))
+      if (! file.exists(dirname(logfile))) dir.create(dirname(logfile))
+      message("Find logging in ", logfile)
+      if (isTRUE(runEnv$qos == "auto")) {
+        sq <- system(paste0("squeue -u ", Sys.info()[["user"]], " -o '%q %j'"), intern = TRUE)
+        runEnv$qos <- if (is.null(attr(sq, "status")) && sum(grepl("^priority ", sq)) < 4) "priority" else "short"
+      }
+      slurmOptions <- combine_slurmConfig(paste0("--qos=", runEnv$qos, " --job-name=", fullrunname, " --output=", logfile,
+        " --open-mode=append --mail-type=END --comment=REMIND-MAgPIE --tasks-per-node=", runEnv$numberOfTasks,
+        if (runEnv$numberOfTasks == 1) " --mem=8000"), runEnv$sbatch)
+      slurmCommand <- paste0("sbatch ", slurmOptions, " --wrap=\"Rscript start_coupled.R coupled_config=", Rdatafile, "\"")
+      message(slurmCommand)
+      exitCode <- system(slurmCommand)
+      if (0 < exitCode) {
+        errorsfound <- errorsfound + 1
+        message("sbatch command failed, check logs.")
       }
     }
   }

tests/testthat/test_01-readDefaultConfig.R

@@ -2,4 +2,7 @@ test_that("readDefaultConfig works", {
   expect_no_warning(cfg <- gms::readDefaultConfig("../.."))
   # make sure that no identical names are used to guarantee unique matching of scenario_config data
   expect_identical(intersect(names(cfg), names(cfg$gms)), character(0))
+  # make sure there is no cm_test and c_test simultaneously
+  shortnames <- tolower(gsub("^[a-zA-Z][a-zA-Z]?_", "", names(cfg$gms)))
+  expect_equal(shortnames[duplicated(shortnames)], character(0))
 })

tutorials/03_RunningBundleOfRuns.md

@@ -65,7 +65,7 @@ The column `slurmConfig` can be used to specify the slurm configuration, either
 
 Everything in the row after a `#` is interpreted as comment. Best use it as first character in the first column to structure the file. Using `#` elsewhere else can lead to unexpected data losses of the cells that follow in the row. If you want to switch off the use of a column, either temporarily or to add some comments, add a dot before the parameter name, which then may read `.cm_startyear` and is then ignored.
 
-Before you start the runs, you can test whether the right runs would be started and find all necessary gdx files. This also writes the `.Rdata` files in the REMIND main folder:
+Before you start the runs, you can test whether the right runs would be started and find all necessary gdx files:
 ```bash
 Rscript start.R --test config/scenario_config_XYZ.csv
 ```