[Pkg-bazaar-commits] ./bzr/unstable r438: - Avoid calling Inventory.iter_entries() when finding modified

[Martin Pool]

------------------------------------------------------------
revno: 438
committer: Martin Pool <mbp at sourcefrog.net>
timestamp: Tue 2005-05-10 16:54:12 +1000
message:
  - Avoid calling Inventory.iter_entries() when finding modified
    files.  Just calculate path for files known to be changed.
  - update_cache optionally takes inventory to avoid reading it twice.
modified:
  bzrlib/commands.py
  bzrlib/statcache.py
-------------- next part --------------
=== modified file 'bzrlib/commands.py'
--- a/bzrlib/commands.py	2005-05-10 06:48:34 +0000
+++ b/bzrlib/commands.py	2005-05-10 06:54:12 +0000
@@ -485,17 +485,23 @@
     def run(self):
         import statcache
         b = Branch('.')
-        sc = statcache.update_cache(b)
+        inv = b.read_working_inventory()
+        sc = statcache.update_cache(b, inv)
         basis = b.basis_tree()
         basis_inv = basis.inventory
-        for path, ie in basis_inv.iter_entries():
-            if ie.kind != 'file':
-                continue
-            cacheentry = sc.get(ie.file_id)
-            if not cacheentry:
-                # deleted
-                continue
+        
+        # We used to do this through iter_entries(), but that's slow
+        # when most of the files are unmodified, as is usually the
+        # case.  So instead we iterate by inventory entry, and only
+        # calculate paths as necessary.
+
+        for file_id in basis_inv:
+            cacheentry = sc.get(file_id)
+            if not cacheentry:                 # deleted
+                continue
+            ie = basis_inv[file_id]
             if cacheentry[statcache.SC_SHA1] != ie.text_sha1:
+                path = inv.id2path(file_id)
                 print path
                 
         

=== modified file 'bzrlib/statcache.py'
--- a/bzrlib/statcache.py	2005-05-10 06:48:34 +0000
+++ b/bzrlib/statcache.py	2005-05-10 06:54:12 +0000
@@ -127,7 +127,7 @@
     
 
 
-def update_cache(branch, flush=False):
+def update_cache(branch, inv=None, flush=False):
     """Update and return the cache for the branch.
 
     The returned cache may contain entries that have not been written
@@ -141,11 +141,13 @@
     # We don't directly know the inum of the files of course but we do
     # know where they were last sighted, so we can sort by that.
 
+    assert isinstance(flush, bool)
     if flush:
         cache = {}
     else:
         cache = load_cache(branch)
-    inv = branch.read_working_inventory()
+    if inv == None:
+        inv = branch.read_working_inventory()
     return _update_cache_from_list(branch, cache, _files_from_inventory(inv))