filectx: use ctx.size comparisons to speed up ctx.cmp

filectx: use ctx.size comparisons to speed up ctx.cmp

Comparing sizes is cheaper than comparing file contents, as it does not
involve reading the file on disk or from the filelog.

It is however not always possible: some extensions, or encode filters,
change data when extracting it to the working directory.

--- a/hgext/keyword.py	Sun Oct 10 18:58:45 2010 +0200
+++ b/hgext/keyword.py	Tue Jul 27 23:07:30 2010 +0900
@@ -82,7 +82,7 @@
 {desc}" expands to the first line of the changeset description.
 '''
 
-from mercurial import commands, cmdutil, dispatch, filelog, extensions
+from mercurial import commands, context, cmdutil, dispatch, filelog, extensions
 from mercurial import localrepo, match, patch, templatefilters, templater, util
 from mercurial.hgweb import webcommands
 from mercurial.i18n import _
@@ -586,6 +586,12 @@
 
     repo.__class__ = kwrepo
 
+    def kwfilectx_cmp(orig, self, fctx):
+        # keyword affects data size, comparing wdir and filelog size does
+        # not make sense
+        return self._filelog.cmp(self._filenode, fctx.data())
+    extensions.wrapfunction(context.filectx, 'cmp', kwfilectx_cmp)
+
     extensions.wrapfunction(patch.patchfile, '__init__', kwpatchfile_init)
     extensions.wrapfunction(patch, 'diff', kw_diff)
     extensions.wrapfunction(cmdutil, 'copy', kw_copy)

--- a/mercurial/context.py	Sun Oct 10 18:58:45 2010 +0200
+++ b/mercurial/context.py	Tue Jul 27 23:07:30 2010 +0900
@@ -357,6 +357,9 @@
 
         returns True if different than fctx.
         """
+        if not self._repo._encodefilterpats and self.size() != fctx.size():
+            return True
+
         return self._filelog.cmp(self._filenode, fctx.data())
 
     def renamed(self):