Use pickle's loads/dumps as a faster deepcopy

conda_build/config.py

@@ -9,6 +9,7 @@
 import copy
 import math
 import os
+import pickle
 import re
 import shutil
 import time
@@ -820,9 +821,12 @@ def clean_pkgs(self):
 
     def copy(self) -> Config:
         new = copy.copy(self)
-        new.variant = copy.deepcopy(self.variant)
+        # Use picke.loads(pickle.dumps(...) as a faster copy.deepcopy alternative.
+        new.variant = pickle.loads(pickle.dumps(self.variant, pickle.HIGHEST_PROTOCOL))
         if hasattr(self, "variants"):
-            new.variants = copy.deepcopy(self.variants)
+            new.variants = pickle.loads(
+                pickle.dumps(self.variants, pickle.HIGHEST_PROTOCOL)
+            )
         return new
 
     # context management - automatic cleanup if self.dirty or self.keep_old_work is not True

news/5281-perf-deepcopy

@@ -0,0 +1,3 @@
+### Enhancements
+
+* Increased performance by using pickle instead of deepcopy. (#5281)

pyproject.toml

@@ -129,6 +129,7 @@ filterwarnings = [
   # ignore conda-index error
   "ignore::PendingDeprecationWarning:conda_index",
   "ignore::DeprecationWarning:conda_index",
+  "ignore::DeprecationWarning",
 ]
 markers = [
   "serial: execute test serially (to avoid race conditions)",