Address cran patch i792

.Rbuildignore

@@ -14,6 +14,7 @@
 LICENSE
 ^cran-comments\.md$
 ^revdep$
+^.keras$
 
 ^logos$
 ^paper$
@@ -54,3 +55,4 @@ LICENSE
 
 # flint files
 ^flint$
+^\.claude$

.github/workflows/R-CMD-check.yaml

@@ -52,8 +52,8 @@ jobs:
         id: get-date
         shell: bash
         run: |
-          echo "::set-output name=year-week::$(date -u "+%Y-%U")"
-          echo "::set-output name=date::$(date -u "+%F")"
+          echo "year-week=$(date -u "+%Y-%U")" >> $GITHUB_OUTPUT
+          echo "date=$(date -u "+%F")" >> $GITHUB_OUTPUT
 
       - name: Restore R package cache
         uses: actions/cache@v4

.gitignore

@@ -21,3 +21,4 @@ revdep/checks.noindex
 revdep/library.noindex
 
 testlog*.md
+.claude

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: greta
-Title: Simple and Scalable Statistical Modelling in R
-Version: 0.5.0.9000
+Title: Simple and 'Scalable' Statistical Modelling in R
+Version: 0.5.1
 Authors@R: c(
     person("Nick", "Golding", , "nick.golding.research@gmail.com", role = "aut",
            comment = c(ORCID = "0000-0001-8916-5570")),
@@ -21,11 +21,11 @@ Authors@R: c(
     person("Jian", "Yen", role = "ctb")
   )
 Description: Write statistical models in R and fit them by MCMC and
-    optimisation on CPUs and GPUs, using Google 'TensorFlow'.  greta lets
+    optimisation on 'CPUs' and 'GPUs', using Google 'TensorFlow'.  'greta' lets
     you write your own model like in BUGS, JAGS and Stan, except that you
     write models right in R, it scales well to massive datasets, and it’s
     easy to extend and build on.  See the website for more information,
-    including tutorials, examples, package documentation, and the greta
+    including tutorials, examples, package documentation, and the 'greta'
     forum.
 License: Apache License 2.0
 URL: https://greta-stats.org, https://github.com/greta-dev/greta
@@ -44,7 +44,7 @@ Imports:
     parallelly (>= 1.29.0),
     progress (>= 1.2.0),
     R6,
-    reticulate (>= 1.19.0),
+    reticulate (>= 1.43.0),
     rlang,
     tensorflow (>= 2.16.0),
     tools,
@@ -84,7 +84,7 @@ Config/testthat/edition: 3
 Encoding: UTF-8
 Language: en-US
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.3.2
+RoxygenNote: 7.3.3
 SystemRequirements: Python (>= 3.7.0) with header files and shared
     library; TensorFlow (>= v2.0.0; https://www.tensorflow.org/); TensorFlow
     Probability (v0.8.0; https://www.tensorflow.org/probability/)

NAMESPACE

@@ -200,6 +200,7 @@ export(greta_create_conda_env)
 export(greta_deps_receipt)
 export(greta_deps_spec)
 export(greta_install_miniconda)
+export(greta_list_py_modules)
 export(greta_notes_tf_num_error)
 export(greta_set_install_logfile)
 export(greta_sitrep)
@@ -285,7 +286,6 @@ importFrom(coda,as.mcmc.list)
 importFrom(coda,mcmc)
 importFrom(coda,mcmc.list)
 importFrom(coda,thin)
-importFrom(future,availableCores)
 importFrom(future,future)
 importFrom(future,nbrOfWorkers)
 importFrom(future,plan)

NEWS.md

@@ -1,11 +1,16 @@
-# greta (development version)
+# greta 0.5.1
 
 ## Changes
 
 - `log.greta_array()` function warns if user uses the `base` arg, as it was unused, (#597).
 - Add warmup information to MCMC print method (#652, resolved by #755).
 - Add more options to level of detail in `greta_sitrep()` with "verbosity" argument. There are three levels, "minimal" (default), "detailed", and "quiet". (#612, resolved by #679).
 - Use `.batch_size` instead of `batch_size` internally, to avoid rare name clash errors (#634).
+- Resolve issues with Tensorflow version in DESCRIPTION (no longer can specify == 2.16.0, must be >= 2.16.0).
+- When the number of cores requested exceeds the number of cores detected, then the number of cores detected will be used.
+- Ensure that `n_cores` arg defaults to 2 cores. Similarly, `chains` defaults to two chains. 
+- Minor internal changes for handling versions of python modules.
+- Added `greta_list_py_modules()` as a function to show the Python packages installed in a specific TF2 environment.
 
 # greta 0.5.0
 

R/checkers.R

@@ -755,16 +755,14 @@ check_cum_op <- function(x, call = rlang::caller_env()) {
   }
 }
 
-
-#' @importFrom future availableCores
 check_n_cores <- function(n_cores, samplers, plan_is) {
-  # if the plan is remote, and the user hasn't specificed the number of cores,
-  # leave it as all of them
+  # if the plan is remote, and the user hasn't specified the number of cores,
+  # still only use 2
   if (is.null(n_cores) & !plan_is$local) {
-    return(NULL)
+    return(2L)
   }
 
-  n_cores_detected <- future::availableCores()
+  n_cores_detected <- parallelly::availableCores()
   allowed_n_cores <- seq_len(n_cores_detected)
 
   cores_exceed_available <- !is.null(n_cores) && !n_cores %in% allowed_n_cores
@@ -773,19 +771,20 @@ check_n_cores <- function(n_cores, samplers, plan_is) {
 
     cli::cli_warn(
       message = "{n_cores} cores were requested, but only {n_cores_detected} \\
-      are available."
+      are available. Setting number of cores to {n_cores_detected}."
     )
 
-    n_cores <- NULL
+    return(as.integer(n_cores_detected))
   }
 
-  # if n_cores isn't user-specified, set it so there's no clash between samplers
-  n_cores <- n_cores %||% floor(n_cores_detected / samplers)
-
-  # make sure there's at least 1
-  n_cores <- max(n_cores, 1)
+  # # if n_cores isn't user-specified, set it so there's no clash between samplers
+  # n_cores <- n_cores %||% floor(n_cores_detected / samplers)
+  # # make sure there's at least 1
+  # n_cores <- max(n_cores, 1)
 
-  as.integer(n_cores)
+  # if n_cores isn't specified, make sure it is set to 2 by default
+  # Resolves #796
+  n_cores %||% 2L
 }
 
 check_positive_integer <- function(x, name = "", call = rlang::caller_env()) {

R/greta-sitrep.R

@@ -6,12 +6,12 @@
 #' @param verbosity character. How verbose the output of the situation report.
 #'   Possible options: "minimal" (default), "detailed", and "quiet". "Minimal"
 #'   provides just information in python version, tensorflow version,
-#'   tensorflow proability, and whether greta conda environment is available.
+#'   tensorflow probability, and whether greta conda environment is available.
 #'   "Quiet" presents no information, but prepares greta to be used. "Detailed"
 #'   gives information on the version and path for R, greta, python,
 #'   tensorflow, tensorflow probability, the greta conda environment, and a
 #'   statement on greta usability.
-#' @return Message on greta situation report. See "verbsoity" parameter details
+#' @return Message on greta situation report. See "verbosity" parameter details
 #'   above for more information.
 #' @export
 #'
@@ -63,7 +63,8 @@ detailed_sitrep <- function() {
   check_if_greta_conda_env_available()
   conda_env_path <- greta_conda_env_path()
   cli::cli_ul("path: {.path {conda_env_path}}")
-  conda_modules <- conda_list_env_modules()
+
+  conda_modules <- greta_list_py_modules()
 
   tf_in_conda <- nzchar(grep(
     "^(tensorflow)(\\s|$)",
@@ -97,12 +98,75 @@ detailed_sitrep <- function() {
   )
 }
 
+#' List Python modules installed in greta env
+#'
+#' @returns matrix/data frame of Python modules that are installed in the greta environment - showing the name, version, build, and install channel.
+#'
+#' @export
+greta_list_py_modules <- function() {
+  conda_modules <- tryCatch(
+    expr = {
+      # This will find conda whether it's on PATH or installed by reticulate
+      conda_bin <- reticulate::conda_binary()
+
+      system2(
+        conda_bin,
+        args = c("list", "-n", "greta-env-tf2"),
+        stdout = TRUE,
+        stderr = TRUE
+      )
+    },
+    error = function(e) {
+      cli::cli_ul(
+        c(
+          "Encountered an error in running:",
+          "{.code conda list -n greta-env-tf2}",
+          "x" = "{.code {e$message}}",
+          "!" = "conda may not be installed. Try {.code reticulate::install_miniconda()}"
+        )
+      )
+      return(NULL)
+    }
+  )
+
+  conda_modules
+}
+
+conda_modules <- tryCatch(
+  expr = {
+    # This will find conda whether it's on PATH or installed by reticulate
+    conda_bin <- reticulate::conda_binary()
+
+    system2(
+      conda_bin,
+      args = c("list", "-n", "greta-env-tf2"),
+      stdout = TRUE,
+      stderr = TRUE
+    )
+  },
+  error = function(e) {
+    cli::cli_ul(
+      c(
+        "Encountered an error in running:",
+        "{.code conda list -n greta-env-tf2}",
+        "x" = "{.code {e$message}}",
+        "!" = "conda may not be installed. Try {.code reticulate::install_miniconda()}"
+      )
+    )
+    return(NULL)
+  }
+)
+
 quiet_sitrep <- function() {
   suppressMessages(check_greta_ready_to_use())
 }
 
 conda_list_env_modules <- function() {
-  system(paste("conda list -n", "greta-env-tf2"), intern = TRUE)
+  system(
+    paste("conda list -n", "greta-env-tf2"),
+    intern = TRUE,
+    ignore.stderr = TRUE
+  )
 }
 
 
@@ -286,9 +350,7 @@ have_tfp <- function() {
   is_tfp_available <- py_module_available("tensorflow_probability")
 
   if (is_tfp_available) {
-    pkg <- reticulate::import("pkg_resources")
-    tfp_version <- pkg$get_distribution("tensorflow_probability")$version
-    is_tfp_available <- utils::compareVersion("0.15.0", tfp_version) <= 0
+    is_tfp_available <- utils::compareVersion("0.15.0", tfp$`__version__`) <= 0
   }
 
   return(is_tfp_available)

R/greta_model_class.R

@@ -141,7 +141,7 @@ print.greta_model <- function(x, ...) {
 #' @details The plot method produces a visual representation of the defined
 #'   model. It uses the `DiagrammeR` package, which must be installed
 #'   first. Here's a key to the plots:
-#'   \if{html}{\figure{plotlegend.png}{options: width="100\%"}}
+#'   \if{html}{\figure{plotlegend.png}{options: width=100}}
 #'   \if{latex}{\figure{plotlegend.pdf}{options: width=7cm}}
 #'
 #' @return `plot` - a [DiagrammeR::grViz()]

R/inference.R

@@ -29,9 +29,9 @@ greta_stash$numerical_messages <- c(
 #'   rest are discarded
 #' @param warmup number of samples to spend warming up the mcmc sampler (moving
 #'   chains toward the highest density area and tuning sampler hyperparameters).
-#' @param chains number of MCMC chains to run
+#' @param chains number of MCMC chains to run. Default is 2. We recommend using more chains as this helps improve convergence. However the number of chains specified can increase the CPU load, so we have to set a lower default value.
 #' @param n_cores the maximum number of CPU cores used by each sampler (see
-#'   details).
+#'   details). If NULL (default), it sets them to 2 cores.
 #' @param verbose whether to print progress information to the console
 #' @param pb_update how regularly to update the progress bar (in iterations).
 #'   If `pb_update` is less than or equal to `thin`, it will be set
@@ -94,9 +94,9 @@ greta_stash$numerical_messages <- c(
 #'   schedulers.
 #'
 #'   If `n_cores = NULL` and mcmc samplers are being run sequentially, each
-#'   sampler will be allowed to use all CPU cores (possibly to compute multiple
-#'   chains sequentially). If samplers are being run in parallel with the
-#'   `future` package, `n_cores` will be set so that `n_cores *
+#'   sampler will be allowed to use only 2 CPU cores (possibly to compute
+#'   multiple chains sequentially). If samplers are being run in parallel with
+#'   the `future` package, `n_cores` will be set so that `n_cores *
 #'   [future::nbrOfWorkers]` is less than the number
 #'   of CPU cores.
 #'
@@ -109,7 +109,7 @@ greta_stash$numerical_messages <- c(
 #'   memory usage.
 #'
 #' @note to set a seed with MCMC you can use [set.seed()], or
-#'   [tensorflow::set_random_seed()]. They both given identical results. See
+#'   [tensorflow::set_random_seed()]. They both give identical results. See
 #'   examples below.
 #'
 #' @return `mcmc`, `stashed_samples` & `extra_samples` - a
@@ -212,7 +212,7 @@ mcmc <- function(
   n_samples = 1000,
   thin = 1,
   warmup = 1000,
-  chains = 4,
+  chains = 2,
   n_cores = NULL,
   verbose = TRUE,
   pb_update = 50,

R/zzz.R

@@ -4,6 +4,14 @@ tf <- tfp <- NULL
 .onLoad <- function(libname, pkgname) {
   # nolint
 
+  # resolve issue with .keras directory
+  Sys.setenv(
+    "KERAS_HOME" = normalizePath(
+      tools::R_user_dir("greta", "cache"),
+      mustWork = FALSE
+    )
+  )
+
   # silence TF's CPU instructions message
   Sys.setenv(TF_CPP_MIN_LOG_LEVEL = 2)
 
@@ -42,5 +50,4 @@ tf <- tfp <- NULL
 
   # warn if TF version is bad
   # check_tf_version("startup")
-
 }