wimlib_iterate_dir_tree(): don't checksum unhashed blobs

wimlib_iterate_dir_tree() on a modified-but-not-committed image is very
slow because it checksums all unhashed blobs.  This was originally
implemented by commit 681faad85f73 ("wimlib_iterate_dir_tree(): checksum
unhashed blobs"), presumably to make the sha1_hash field always valid.
However, I can't remember a real use case for this.  The current
behavior is causing problems, so let's just revert it and update the
documentation accordingly.

Reported at https://wimlib.net/forums/viewtopic.php?f=1&t=572

diff --git a/include/wimlib.h b/include/wimlib.h
index c032429d6e9d1863fd72f2273f7bcee55b5f0e86..f5c9e0e842666b7651b3882b8e9972f7dda62208 100644 (file)
--- a/include/wimlib.h
+++ b/include/wimlib.h
@@ -1399,20 +1399,23 @@ struct wimlib_wim_info {
  *    sha1_hash.  This case can only occur with wimlib_iterate_dir_tree(), never
  *    wimlib_iterate_lookup_table().
  *
- * 2. Otherwise we know the sha1_hash, the uncompressed_size, the
- *    reference_count, and the is_metadata flag.  In addition:
+ * 2. Otherwise we know the the uncompressed_size, the reference_count, and the
+ *    is_metadata flag.  In addition:
  *
  *    A. If the blob is located in a non-solid WIM resource, then we also know
- *       the compressed_size and offset.
+ *       the sha1_hash, compressed_size, and offset.
  *
  *    B. If the blob is located in a solid WIM resource, then we also know the
- *       offset, raw_resource_offset_in_wim, raw_resource_compressed_size, and
- *       raw_resource_uncompressed_size.  But the "offset" is actually the
- *       offset in the uncompressed solid resource rather than the offset from
- *       the beginning of the WIM file.
+ *       sha1_hash, offset, raw_resource_offset_in_wim,
+ *       raw_resource_compressed_size, and raw_resource_uncompressed_size.  But
+ *       the "offset" is actually the offset in the uncompressed solid resource
+ *       rather than the offset from the beginning of the WIM file.
  *
- *    C. If the blob is *not* located in any type of WIM resource, then we don't
- *       know any additional information.
+ *    C. If the blob is *not* located in any type of WIM resource, for example
+ *       if it's in a external file that was scanned by wimlib_add_image(), then
+ *       we usually won't know any more information.  The sha1_hash might be
+ *       known, and prior to wimlib v1.13.6 it always was; however, in wimlib
+ *       v1.13.6 and later, the sha1_hash might not be known in this case.
  *
  * Unknown or irrelevant fields are left zeroed.
  */
@@ -1432,7 +1435,8 @@ struct wimlib_resource_entry {
         * of this blob within that solid resource when uncompressed.  */
        uint64_t offset;
 
-       /** The SHA-1 message digest of the blob's uncompressed contents.  */
+       /** If this blob is located in a WIM resource, then this is the SHA-1
+        * message digest of the blob's uncompressed contents.  */
        uint8_t sha1_hash[20];
 
        /** If this blob is located in a WIM resource, then this is the part

diff --git a/src/iterate_dir.c b/src/iterate_dir.c
index 940fc24b8e2c96d3635c6eb31e4d4b8407a1ec34..b6fdb0320419d4d3de6d5fb7d3823109ce98f661 100644 (file)
--- a/src/iterate_dir.c
+++ b/src/iterate_dir.c
@@ -269,11 +269,6 @@ wimlib_iterate_dir_tree(WIMStruct *wim, int image, const tchar *_path,
        path = canonicalize_wim_path(_path);
        if (path == NULL)
                return WIMLIB_ERR_NOMEM;
-
-       ret = wim_checksum_unhashed_blobs(wim);
-       if (ret)
-               return ret;
-
        struct image_iterate_dir_tree_ctx ctx = {
                .path = path,
                .flags = flags,