[Code scan] Avoid mutable defaults in Task and Resources

[njzjz]

Found by a Codex global repository scan of deepmodeling/dpdispatcher at commit 98a9e08.

Problem
Task and Resources constructors use mutable objects as default argument values and store caller-provided lists/dicts directly.

Relevant code

dpdispatcher/dpdispatcher/submission.py

Lines 567 to 579 in 98a9e08

	def __init__(
	self,
	command,
	task_work_path,
	forward_files=[],
	backward_files=[],
	outlog="log",
	errlog="err",
	):
	self.command = command
	self.task_work_path = task_work_path
	self.forward_files = forward_files
	self.backward_files = backward_files

dpdispatcher/dpdispatcher/submission.py

Lines 1028 to 1065 in 98a9e08

	def __init__(
	self,
	number_node,
	cpu_per_node,
	gpu_per_node,
	queue_name,
	group_size,
	*,
	custom_flags=[],
	strategy=default_strategy,
	para_deg=1,
	module_unload_list=[],
	module_purge=False,
	module_list=[],
	source_list=[],
	envs={},
	prepend_script=[],
	append_script=[],
	wait_time=0,
	**kwargs,
	):
	self.number_node = number_node
	self.cpu_per_node = cpu_per_node
	self.gpu_per_node = gpu_per_node
	self.queue_name = queue_name
	self.group_size = group_size
	# self.extra_specification = extra_specification
	self.custom_flags = custom_flags
	self.strategy = strategy
	self.para_deg = para_deg
	self.module_purge = module_purge
	self.module_unload_list = module_unload_list
	self.module_list = module_list
	self.source_list = source_list
	self.envs = envs
	self.prepend_script = prepend_script
	self.append_script = append_script

dpdispatcher/dpdispatcher/utils/dpcloudserver/zip_file.py

Lines 10 to 16 in 98a9e08

	def zip_file_list(root_path, zip_filename, file_list=[]):
	out_zip_file = os.path.join(root_path, zip_filename)
	# print('debug: file_list', file_list)
	zip_obj = ZipFile(out_zip_file, "w")
	for f in file_list:
	matched_files = os.path.join(root_path, f)
	for ii in glob.glob(matched_files):

Impact
Mutating one object's forward_files, custom_flags, strategy, envs, or similar collection can affect later objects in the same Python process. Caller-owned lists and dicts can also be mutated after construction and change object behavior unexpectedly.

Suggested fix
Use None defaults and allocate fresh lists/dicts inside __init__. Copy caller-provided mutable collections before storing them.