Better default configuration of Galaxy test data directory.

config/galaxy.yml.sample

@@ -1167,6 +1167,15 @@ galaxy:
   # Set maximum size of ngrams
   #tool_ngram_maxsize: 4
 
+  # Set tool test data directory. The test framework sets this value to
+  # 'test-data,https://github.com/galaxyproject/galaxy-test-data.git'
+  # which will cause Galaxy to clone down extra test data on the fly for
+  # certain tools distributed with Galaxy but this is likely not
+  # appropriate for production systems. Instead one can simply clone
+  # that repository directly and specify a path here instead of a Git
+  # HTTP repository.
+  #tool_test_data_directories: test-data
+
   # Galaxy encodes various internal values when these values will be
   # output in some format (for example, in a URL or cookie).  You should
   # set a key to be used by the algorithm that encodes and decodes these

doc/source/admin/galaxy_options.rst

@@ -2439,6 +2439,22 @@
 :Type: int
 
 
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+``tool_test_data_directories``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+:Description:
+    Set tool test data directory. The test framework sets this value
+    to 'test-data,https://github.com/galaxyproject/galaxy-test-
+    data.git' which will cause Galaxy to clone down extra test data on
+    the fly for certain tools distributed with Galaxy but this is
+    likely not appropriate for production systems. Instead one can
+    simply clone that repository directly and specify a path here
+    instead of a Git HTTP repository.
+:Default: ``test-data``
+:Type: str
+
+
 ~~~~~~~~~~~~~
 ``id_secret``
 ~~~~~~~~~~~~~

lib/galaxy/app.py

@@ -95,7 +95,7 @@ class UniverseApplication(config.ConfiguresGalaxyMixin):
         self.dataset_collections_service = DatasetCollectionManager(self)
         self.history_manager = HistoryManager(self)
         self.dependency_resolvers_view = DependencyResolversView(self)
-        self.test_data_resolver = test_data.TestDataResolver()
+        self.test_data_resolver = test_data.TestDataResolver(file_dirs=getattr(self.config, "tool_test_data_directories", None))
 
         # Tool Data Tables
         self._configure_tool_data_tables(from_shed_config=False)

lib/galaxy/config.py

@@ -462,6 +462,8 @@ class Configuration(object):
         self.tool_enable_ngram_search = kwargs.get("tool_enable_ngram_search", False)
         self.tool_ngram_minsize = kwargs.get("tool_ngram_minsize", 3)
         self.tool_ngram_maxsize = kwargs.get("tool_ngram_maxsize", 4)
+        default_tool_test_data_directories = os.environ.get("GALAXY_TEST_FILE_DIR", resolve_path("test-data", self.root))
+        self.tool_test_data_directories = kwargs.get("tool_test_data_directories", default_tool_test_data_directories)
         # Location for tool dependencies.
         use_tool_dependencies, tool_dependency_dir, use_cached_dependency_manager, tool_dependency_cache_dir, precache_dependencies = \
             parse_dependency_options(kwargs, self.root, self.dependency_resolvers_config_file)

lib/galaxy/tools/verify/test_data.py

@@ -27,8 +27,9 @@ LIST_SEP = re.compile("\s*,\s*")
 
 class TestDataResolver(object):
 
-    def __init__(self, env_var='GALAXY_TEST_FILE_DIR', environ=os.environ):
-        file_dirs = environ.get(env_var, None)
+    def __init__(self, file_dirs=None, env_var='GALAXY_TEST_FILE_DIR', environ=os.environ):
+        if file_dirs is None:
+            file_dirs = environ.get(env_var, None)
         if file_dirs is None:
             file_dirs = "test-data,https://github.com/galaxyproject/galaxy-test-data.git"
         if file_dirs:

lib/galaxy/webapps/galaxy/config_schema.yml

@@ -1817,6 +1817,18 @@ mapping:
         desc: |
           Set maximum size of ngrams
 
+      tool_test_data_directories:
+        type: str
+        default: 'test-data'
+        required: false
+        desc: |
+          Set tool test data directory. The test framework sets this value to
+          'test-data,https://github.com/galaxyproject/galaxy-test-data.git' which will
+          cause Galaxy to clone down extra test data on the fly for certain tools
+          distributed with Galaxy but this is likely not appropriate for production systems.
+          Instead one can simply clone that repository directly and specify a path here
+          instead of a Git HTTP repository.
+
       id_secret:
         type: str
         default: USING THE DEFAULT IS NOT SECURE!