feat(build-all): add trim_pkgcache_metadata helper to bound pkgcache _metadata

local/r-minor-helpers.R

@@ -31,3 +31,42 @@ parse_build_args <- function(args) {
     ncpus = as.integer(pos[3L])
   )
 }
+
+# Bound the {pkgcache} metadata dir, which otherwise grows without limit: the
+# "patched" repo mints a new content hash on every PACKAGES change, so each build
+# writes a fresh ~70 MB _metadata/pkgs-<hash>.rds (+ patched-<hash>/) that is
+# never reused. Keep the `keep` newest entries by mtime; only remove entries
+# older than `min_age_secs`, so a concurrent split-job's in-flight files are
+# never deleted (each build uses a unique hash, so aged entries are
+# unreferenced). Stable repo dirs (CRAN-*, BioC*, INLA-*) and pkg/ downloads are
+# not matched and thus preserved. Returns the number of entries removed.
+trim_pkgcache_metadata <- function(cache_dir = Sys.getenv("R_PKG_CACHE_DIR"),
+                                   keep = 20L,
+                                   min_age_secs = 600) {
+  meta <- file.path(cache_dir, "R", "pkgcache", "_metadata")
+  if (!nzchar(cache_dir) || !dir.exists(meta)) {
+    return(0L)
+  }
+  entries <- c(
+    Sys.glob(file.path(meta, "patched-*")),
+    Sys.glob(file.path(meta, "pkgs-*.rds"))
+  )
+  if (length(entries) == 0L) {
+    return(0L)
+  }
+  info <- file.info(entries)
+  order_new_first <- order(info$mtime, decreasing = TRUE)
+  ranked <- entries[order_new_first]
+  ranked_mtime <- info$mtime[order_new_first]
+  if (length(ranked) <= keep) {
+    return(0L)
+  }
+  candidates <- ranked[(keep + 1L):length(ranked)]
+  candidate_age <- as.numeric(Sys.time()) - as.numeric(ranked_mtime[(keep + 1L):length(ranked)])
+  removable <- candidates[candidate_age >= min_age_secs]
+  if (length(removable) == 0L) {
+    return(0L)
+  }
+  unlink(removable, recursive = TRUE, force = TRUE)
+  length(removable)
+}