Author: cmpilato
Revision: 1101728
Modified property: svn:log

Modified: svn:log at Wed May 11 03:48:36 2011
------------------------------------------------------------------------------
--- svn:log (original)
+++ svn:log Wed May 11 03:48:36 2011
@@ -1,20 +1,23 @@
-* subversion/libsvn_wc/adm_ops.c
-  (revert_restore): Ok, thinking this over at 4 AM. I don't think we are
-    willing to accept the performance regression of r1101473 in svn revert -R.
-    So I accept that we don't have to revert that are not strictly as checked
-    out, but are still unmodified according to status.
+Ok, thinking this over at 4 AM. I don't think we are willing to accept
+the performance regression of r1101473 in svn revert -R.  So I accept
+that we don't have to revert that are not strictly as checked out, but
+are still unmodified according to status.
+
+In most cases this reverts to the behavior of before r1101473, as
+before that we accepted recorded information as 'the file is not
+modified'. But I'm introducing the change now that when the file might
+be modified, we do the less strict comparison.
 
-    In most cases this reverts to the behavior of before r1101473, as before
-    that we accepted recorded information as 'the file is not modified'. But
-    I'm introducing the change now that when the file might be modified, we
-    do the less strict comparison.
+The only way to have both the performance and the strict comparison
+would be to store another boolean with the recorded information to
+tell to what level we performed the recorded comparison.  (And should
+we then cache the strict or the not strict?)
 
-    The only way to have both the performance and the strict comparison would
-    be to store another boolean with the recorded information to tell to what
-    level we performed the recorded comparison.
-    (And should we then cache the strict or the not strict?)
+[Note from the future: This reverts 'svn revert' to the 1.6.x behavior
+ of using the recorded information as hint.  But this breaks one of
+ the revert tests because that tests assumes that 'svn revert' notices
+ keyword expansion changes.  More details in r1101730.]
 
-    [Note from the future: This reverts revert to the 1.6.x behavior of using
-     the recorded information as hint. But this breaks one of the revert tests
-     because that assumes that revert notices keyword expansion changes.
-     More details in r1101730.]
+* subversion/libsvn_wc/adm_ops.c
+  (revert_restore): No longer ask svn_wc__internal_file_modified_p()
+    to perform an exact comparison.

Reply via email to