stream-clone: disable gc for `_entries_walk` duration
The number of small container created turn Python in a gc-frenzy that seriously
impact performance.
This significantly boost performance. The following number comes from a large
private repository using perf::stream-locked-section:
base-line: 35.04 seconds
prev-change: 24.51 seconds (-30%)
this-change: 20.88 seconds (-40% from baseline; -15% from previous changes)
#testcases bdiff xdiff
#if xdiff
#require xdiff
$ cat >> $HGRCPATH <<EOF
> [experimental]
> xdiff = true
> EOF
#endif
Test case that makes use of the weakness of patience diff algorithm
$ hg init
>>> open('a', 'wb').write(('\n'.join(list('a' + 'x' * 10 + 'u' + 'x' * 30 + 'a\n'))).encode('ascii')) and None
$ hg commit -m 1 -A a
>>> open('a', 'wb').write(('\n'.join(list('b' + 'x' * 30 + 'u' + 'x' * 10 + 'b\n'))).encode('ascii')) and None
#if xdiff
$ hg diff
diff -r f0aeecb49805 a
--- a/a Thu Jan 01 00:00:00 1970 +0000
+++ b/a Thu Jan 01 00:00:00 1970 +0000
@@ -1,4 +1,4 @@
-a
+b
x
x
x
@@ -9,7 +9,6 @@
x
x
x
-u
x
x
x
@@ -30,6 +29,7 @@
x
x
x
+u
x
x
x
@@ -40,5 +40,5 @@
x
x
x
-a
+b
#else
$ hg diff
diff -r f0aeecb49805 a
--- a/a Thu Jan 01 00:00:00 1970 +0000
+++ b/a Thu Jan 01 00:00:00 1970 +0000
@@ -1,15 +1,4 @@
-a
-x
-x
-x
-x
-x
-x
-x
-x
-x
-x
-u
+b
x
x
x
@@ -40,5 +29,16 @@
x
x
x
-a
+u
+x
+x
+x
+x
+x
+x
+x
+x
+x
+x
+b
#endif