From fb1a4fd150b8a31c8c023451ac1cd59b6794b805 Mon Sep 17 00:00:00 2001
From: Frank <frank@dankware>
Date: Tue, 6 Oct 2026 00:03:58 -0700
Subject: [PATCH] frank-primary: never write the stage index through the raw
 symlink

Run 3 rewrote the staged index THROUGH the relative symlink into the raw
checkpoint: the raw index.json was restamped with the stacked layout
(605 KB smaller), the raw verify then failed 143/144 and the overnight
run died before stacking. Two fixes:

- stack-qwen38-official-fp8.py: index/config writes are now temp+os.replace
  (an atomic swap replaces the symlink itself; the raw tree stays pristine).
- build-frank-primary-rad.sh: the stage's index.json is COPIED like
  config.json; only shards stay relative symlinks.
---
 scripts/build-frank-primary-rad.sh   | 11 ++++++++---
 scripts/stack-qwen38-official-fp8.py | 15 ++++++++++-----
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/scripts/build-frank-primary-rad.sh b/scripts/build-frank-primary-rad.sh
index 895b32c..3fb08ee 100755
--- a/scripts/build-frank-primary-rad.sh
+++ b/scripts/build-frank-primary-rad.sh
@@ -98,13 +98,18 @@ dropped = cfg.pop("quantization_config", None)
 json.dump(cfg, open(dst, "w"), indent=2)
 print(f"staging config.json written (quantization_config dropped: {dropped is not None})")
 PY
-    # Everything else is a RELATIVE symlink — relative so the targets resolve
-    # inside the convert container (the stage's parent is mounted at /models).
+    # The index is COPIED like config.json, never symlinked: the reshape step
+    # rewrites the stage's index, and a symlink would write through to the
+    # raw checkpoint (measured 2026-10-05: the raw index was restamped with
+    # the stacked layout and the raw verify failed 143/144). Shards stay
+    # relative symlinks — relative so the targets resolve inside the convert
+    # container (the stage's parent is mounted at /models).
     for f in "$SRC"/*; do
         name="$(basename "$f")"
-        [ "$name" = "config.json" ] && continue
+        case "$name" in config.json|model.safetensors.index.json) continue ;; esac
         ln -sfn "../Qwen/Qwen3.8-Flash-Next-FP8/$name" "$STAGE/$name"
     done
+    cp "$SRC/model.safetensors.index.json" "$STAGE/"
     echo "staging dir ready: $STAGE ($(ls "$STAGE" | wc -l) entries)"
 fi
 
diff --git a/scripts/stack-qwen38-official-fp8.py b/scripts/stack-qwen38-official-fp8.py
index c5db5d2..3604cc0 100755
--- a/scripts/stack-qwen38-official-fp8.py
+++ b/scripts/stack-qwen38-official-fp8.py
@@ -424,12 +424,17 @@ def main():
     index["weight_map"] = new_wm
     if "metadata" in index and isinstance(index["metadata"], dict):
         index["metadata"]["total_size"] = total
-    with open(index_path, "w") as fh:
-        json.dump(index, fh, indent=2)
-
+    # ATOMIC REPLACE, never open(path, "w"): the stage's index is a symlink
+    # to the RAW checkpoint (staging symlinks everything but config.json) —
+    # open("w") would rewrite the raw file through the link. os.replace
+    # swaps the link itself for a real file; the raw tree stays pristine.
+    for path, payload in ((index_path, json.dumps(index, indent=2)),
+                          (cfg_path, json.dumps(cfg, indent=2))):
+        tmp = path + ".tmp"
+        with open(tmp, "w") as fh:
+            fh.write(payload)
+        os.replace(tmp, path)
     n = set_split(cfg, 1)
-    with open(cfg_path, "w") as fh:
-        json.dump(cfg, fh, indent=2)
     log(f"config: split_ngram_parts=1 written at {n} location(s)")
 
     with open(os.path.join(stage, "ple_scale.txt"), "w") as fh: