From d7e8bed73568fd11fcb40f2def108fc8a8334090 Mon Sep 17 00:00:00 2001
From: bghira <bghira@users.github.com>
Date: Sat, 3 Aug 2024 23:48:20 -0600
Subject: [PATCH] bugfix: correctly set hash_filenames to true or false for an
 initial dataset creation

---
 helpers/data_backend/factory.py | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/helpers/data_backend/factory.py b/helpers/data_backend/factory.py
index 3fda5f7b..7e794f75 100644
--- a/helpers/data_backend/factory.py
+++ b/helpers/data_backend/factory.py
@@ -534,7 +534,6 @@ def configure_multi_databackend(
         info_log(f"Configuring data backend: {backend['id']}")
         # Retrieve some config file overrides for commandline arguments, eg. cropping
         init_backend = init_backend_config(backend, args, accelerator)
-        info_log(f"Configured backend: {init_backend}")
         StateTracker.set_data_backend_config(
             data_backend_id=init_backend["id"],
             config=init_backend["config"],
@@ -730,9 +729,7 @@ def configure_multi_databackend(
             if current_config_version is None:
                 # backwards compatibility for non-versioned config files, so that we do not enable life-changing options.
                 current_config_version = 1
-                init_backend["config"]["hash_filenames"] = False
-            if "hash_filenames" not in prev_config and current_config_version < 2:
-                init_backend["config"]["hash_filenames"] = False
+
             logger.debug(
                 f"Found existing config (version={current_config_version}): {prev_config}"
             )
@@ -886,6 +883,9 @@ def configure_multi_databackend(
         hash_filenames = init_backend.get("config", {}).get(
             "hash_filenames", default_hash_option
         )
+        init_backend["config"]["hash_filenames"] = hash_filenames
+        StateTracker.set_data_backend_config(init_backend["id"], init_backend["config"])
+        logger.debug(f"Hashing filenames: {hash_filenames}")
 
         if "deepfloyd" not in StateTracker.get_args().model_type:
             info_log(f"(id={init_backend['id']}) Creating VAE latent cache.")
@@ -978,6 +978,8 @@ def configure_multi_databackend(
             )
             accelerator.wait_for_everyone()
 
+        info_log(f"Configured backend: {init_backend}")
+
         StateTracker.register_data_backend(init_backend)
         init_backend["metadata_backend"].save_cache()