Benchmark require #7158

0000-bench_require_empty.rb

#!/usr/bin/env ruby
require "benchmark"

in_dir = File.expand_path('input/empty_files', File.dirname(__FILE__))

200.times do |i|
  $LOAD_PATH.push File.expand_path(i.to_s)
end
$LOAD_PATH.push in_dir

Benchmark.bm do |x|
  x.report do
    1000.times do |i|
      require "empty#{i}"
    end
  end
end

puts "LOAD_PATH SIZE = #{$:.size}"
puts "LOADED_FEATURES SIZE = #{$".size}"

0000-bench_require_empty_relative.rb

#!/usr/bin/env ruby
require "benchmark"

in_dir = File.expand_path('input/empty_files', File.dirname(__FILE__))

200.times do |i|
  $LOAD_PATH.push i.to_s
end
$LOAD_PATH.push in_dir

Benchmark.bm do |x|
  x.report do
    1000.times do |i|
      require "empty#{i}"
    end
  end
end

puts "LOAD_PATH SIZE = #{$:.size}"
puts "LOADED_FEATURES SIZE = #{$".size}"

0000-result.md

Best of 5 trials

LOAD_PATH SIZE = 209

LOADED_FEATURES SIZE = 1010

# trunk
ruby 2.0.0dev (2012-10-27 trunk 37341) [x86_64-darwin12.2.0]
       user     system      total        real
   0.620000   0.390000   1.010000 (  1.022564)

# Greg's patch
ruby 2.0.0dev (2012-10-27 trunk 37341) [x86_64-darwin12.2.0]
       user     system      total        real
   0.380000   0.350000   0.730000 (  0.729237)

# Greg's patch + my patch 1 and 2
ruby 2.0.0dev (2012-10-27 trunk 37341) [x86_64-darwin12.2.0]
       user     system      total        real
   0.380000   0.350000   0.730000 (  0.736977)

# Greg's patch + my patch 1, 2 and 3
ruby 2.0.0dev (2012-10-27 trunk 37341) [x86_64-darwin12.2.0]
       user     system      total        real
   0.380000   0.350000   0.730000 (  0.727115)

0000-result2.md

patched contains all my patches.

# trunk bench_require_empty.rb
ruby 2.0.0dev (2012-11-01 trunk 37415) [x86_64-darwin12.2.0]
       user     system      total        real
   0.610000   0.390000   1.000000 (  1.002916)
LOAD_PATH SIZE = 209 
LOADED_FEATURES SIZE = 1010

# patched bench_require_empty.rb
ruby 2.0.0dev (2012-11-01 trunk 37415) [x86_64-darwin12.2.0]
       user     system      total        real
   0.390000   0.350000   0.740000 (  0.735971)
LOAD_PATH SIZE = 209 
LOADED_FEATURES SIZE = 1010


# trunk bench_require_empty_relative.rb
ruby 2.0.0dev (2012-11-01 trunk 37415) [x86_64-darwin12.2.0]
       user     system      total        real
   1.600000   2.470000   4.070000 (  4.068619)
LOAD_PATH SIZE = 209 
LOADED_FEATURES SIZE = 1010

# patched bench_require_empty_relative.rb
ruby 2.0.0dev (2012-11-01 trunk 37415) [x86_64-darwin12.2.0]
       user     system      total        real
   0.420000   0.360000   0.780000 (  0.787250)
LOAD_PATH SIZE = 209 
LOADED_FEATURES SIZE = 1010

0001-Expand-relative-load-path-always.patch

From 1245d796305be72b80a8c623faf3195ee92fb1f3 Mon Sep 17 00:00:00 2001
From: Hiroshi Shirosaki <[email protected]>
Date: Sat, 27 Oct 2012 18:39:01 +0900
Subject: [PATCH 1/3] Expand relative load path always

* load.c (rb_construct_expanded_load_path): add a option for expanding
  only relative load paths.
* load.c (rb_get_expanded_load_path): expand relative load path always.
---
 load.c |   15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/load.c b/load.c
index dfea0ef..b3e316c 100644
--- a/load.c
+++ b/load.c
@@ -34,10 +34,11 @@ rb_get_load_path(void)
 }
 
 static void
-rb_construct_expanded_load_path(void)
+rb_construct_expanded_load_path(int only_relative)
 {
     rb_vm_t *vm = GET_VM();
     VALUE load_path = vm->load_path;
+    VALUE expanded_load_path = vm->expanded_load_path;
     VALUE ary;
     long i;
 
@@ -46,6 +47,10 @@ rb_construct_expanded_load_path(void)
 	VALUE path, as_str, expanded_path;
 	as_str = path = RARRAY_PTR(load_path)[i];
 	StringValue(as_str);
+	if (only_relative && rb_is_absolute_path(StringValuePtr(as_str))) {
+	    rb_ary_push(ary, RARRAY_PTR(expanded_load_path)[i]);
+	    continue;
+	}
 	if (as_str != path)
 	    rb_ary_store(load_path, i, as_str);
 	rb_str_freeze(as_str);
@@ -63,8 +68,12 @@ rb_get_expanded_load_path(void)
 {
     rb_vm_t *vm = GET_VM();
     if (!rb_ary_shared_with_p(vm->load_path_snapshot, vm->load_path)) {
-	/* The load path was modified.	Rebuild the expanded load path. */
-	rb_construct_expanded_load_path();
+	/* The load path was modified. Rebuild the expanded load path. */
+	rb_construct_expanded_load_path(0);
+    }
+    else {
+	/* Expand only relative load path. */
+	rb_construct_expanded_load_path(1);
     }
     return vm->expanded_load_path;
 }
-- 
1.7.10.2 (Apple Git-33)

0002-Cache-expanded-relative-load-path.patch

From f0a7f264e53b64b0493c8de3b2d592bd5003bc0c Mon Sep 17 00:00:00 2001
From: Hiroshi Shirosaki <[email protected]>
Date: Sat, 27 Oct 2012 20:50:41 +0900
Subject: [PATCH 2/3] Cache expanded relative load path

* load.c (rb_construct_expanded_load_path): check relative path.
  We call getcwd() only if load path has relative path, to avoid getcwd() failure.

* load.c (load_path_getcwd): add a function to get current working directory.

* load.c (rb_get_expanded_load_path): check if current working directory is changed
  and rebuild expanded load path cache.

* load.c (Init_load): initialize vm->load_path_cwd.

* test/ruby/test_require.rb (TestRequire#test_require_): add a test for require
  when changeing current working directory.

* vm.c (rb_vm_mark): mark vm->load_path_cwd.

* vm_core.h (rb_vm_struct): add vm->load_path_cwd to store current working directory
  to check if expanded load path cache is invalid.
---
 load.c                    |   33 +++++++++++++++++++++++++++------
 test/ruby/test_require.rb |   22 ++++++++++++++++++++++
 vm.c                      |    1 +
 vm_core.h                 |    1 +
 4 files changed, 51 insertions(+), 6 deletions(-)

diff --git a/load.c b/load.c
index b3e316c..53dc58c 100644
--- a/load.c
+++ b/load.c
@@ -34,7 +34,7 @@ rb_get_load_path(void)
 }
 
 static void
-rb_construct_expanded_load_path(int only_relative)
+rb_construct_expanded_load_path(int only_relative, int *has_relative)
 {
     rb_vm_t *vm = GET_VM();
     VALUE load_path = vm->load_path;
@@ -51,6 +51,8 @@ rb_construct_expanded_load_path(int only_relative)
 	    rb_ary_push(ary, RARRAY_PTR(expanded_load_path)[i]);
 	    continue;
 	}
+	if (!*has_relative && !rb_is_absolute_path(StringValuePtr(as_str)))
+	    *has_relative = 1;
 	if (as_str != path)
 	    rb_ary_store(load_path, i, as_str);
 	rb_str_freeze(as_str);
@@ -64,16 +66,34 @@ rb_construct_expanded_load_path(int only_relative)
 }
 
 static VALUE
+load_path_getcwd(void)
+{
+    char *cwd = my_getcwd();
+    VALUE cwd_str = rb_filesystem_str_new_cstr(cwd);
+    xfree(cwd);
+    return cwd_str;
+}
+
+static VALUE
 rb_get_expanded_load_path(void)
 {
     rb_vm_t *vm = GET_VM();
+
     if (!rb_ary_shared_with_p(vm->load_path_snapshot, vm->load_path)) {
 	/* The load path was modified. Rebuild the expanded load path. */
-	rb_construct_expanded_load_path(0);
-    }
-    else {
-	/* Expand only relative load path. */
-	rb_construct_expanded_load_path(1);
+	int has_relative = 0;
+	rb_construct_expanded_load_path(0, &has_relative);
+	vm->load_path_cwd = has_relative ? load_path_getcwd() : 0;
+    }
+    else if (vm->load_path_cwd) {
+	VALUE cwd = load_path_getcwd();
+	if (!rb_str_equal(vm->load_path_cwd, cwd)) {
+	    /* Current working directory or filesystem encoding was changed.
+	       Expand relative load path again. */
+	    int has_relative = 1;
+	    vm->load_path_cwd = cwd;
+	    rb_construct_expanded_load_path(1, &has_relative);
+	}
     }
     return vm->expanded_load_path;
 }
@@ -971,6 +991,7 @@ Init_load()
     vm->load_path = rb_ary_new();
     vm->expanded_load_path = rb_ary_new();
     vm->load_path_snapshot = rb_ary_new();
+    vm->load_path_cwd = 0;
 
     rb_define_virtual_variable("$\"", get_loaded_features, 0);
     rb_define_virtual_variable("$LOADED_FEATURES", get_loaded_features, 0);
diff --git a/test/ruby/test_require.rb b/test/ruby/test_require.rb
index a7c75d0..8c72571 100644
--- a/test/ruby/test_require.rb
+++ b/test/ruby/test_require.rb
@@ -435,4 +435,26 @@ class TestRequire < Test::Unit::TestCase
     $:.replace(loadpath)
     $".replace(features)
   end
+
+  def test_require_changed_current_dir
+    Dir.mktmpdir {|tmp|
+      Dir.chdir(tmp) {
+        Dir.mkdir("a")
+        Dir.mkdir("b")
+        open(File.join("a", "foo.rb"), "w") {}
+        open(File.join("b", "bar.rb"), "w") {|f|
+          f.puts "p :ok"
+        }
+        assert_in_out_err([], <<-INPUT, %w(:ok), [])
+          $: << "."
+          Dir.chdir("a")
+          require "foo"
+          Dir.chdir("../b")
+          p :ng unless require "bar"
+          Dir.chdir("..")
+          p :ng if require "b/bar"
+        INPUT
+      }
+    }
+  end
 end
diff --git a/vm.c b/vm.c
index 9a8ebef..03cc79d 100644
--- a/vm.c
+++ b/vm.c
@@ -1508,6 +1508,7 @@ rb_vm_mark(void *ptr)
 	RUBY_MARK_UNLESS_NULL(vm->mark_object_ary);
 	RUBY_MARK_UNLESS_NULL(vm->load_path);
 	RUBY_MARK_UNLESS_NULL(vm->load_path_snapshot);
+	RUBY_MARK_UNLESS_NULL(vm->load_path_cwd);
 	RUBY_MARK_UNLESS_NULL(vm->expanded_load_path);
 	RUBY_MARK_UNLESS_NULL(vm->loaded_features);
 	RUBY_MARK_UNLESS_NULL(vm->loaded_features_snapshot);
diff --git a/vm_core.h b/vm_core.h
index b315fe4..3874231 100644
--- a/vm_core.h
+++ b/vm_core.h
@@ -355,6 +355,7 @@ typedef struct rb_vm_struct {
     VALUE top_self;
     VALUE load_path;
     VALUE load_path_snapshot;
+    VALUE load_path_cwd;
     VALUE expanded_load_path;
     VALUE loaded_features;
     VALUE loaded_features_snapshot;
-- 
1.7.10.2 (Apple Git-33)

0003-Use-rb_get_expanded_load_path-in-find-file.patch

From 6522cc9cf8b8ac918e09705342f5ad9d5a5c31f6 Mon Sep 17 00:00:00 2001
From: Hiroshi Shirosaki <[email protected]>
Date: Sat, 27 Oct 2012 21:46:59 +0900
Subject: [PATCH 3/3] Use rb_get_expanded_load_path() in find file

* file.c (rb_find_file_ext_safe): use rb_get_expanded_load_path() to reduce
  expand cost.
* file.c (rb_find_file_safe): ditto.
* internal.h (rb_get_expanded_load_path): add a declaration.
* load.c (rb_get_expanded_load_path): make it accesible from other source
  files.
---
 file.c     |    4 ++--
 internal.h |    1 +
 load.c     |    2 +-
 3 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/file.c b/file.c
index 5951d7b..7b22dcf 100644
--- a/file.c
+++ b/file.c
@@ -5258,7 +5258,7 @@ rb_find_file_ext_safe(VALUE *filep, const char *const *ext, int safe_level)
 	rb_raise(rb_eSecurityError, "loading from non-absolute path %s", f);
     }
 
-    RB_GC_GUARD(load_path) = rb_get_load_path();
+    RB_GC_GUARD(load_path) = rb_get_expanded_load_path();
     if (!load_path) return 0;
 
     fname = rb_str_dup(*filep);
@@ -5323,7 +5323,7 @@ rb_find_file_safe(VALUE path, int safe_level)
 	rb_raise(rb_eSecurityError, "loading from non-absolute path %s", f);
     }
 
-    RB_GC_GUARD(load_path) = rb_get_load_path();
+    RB_GC_GUARD(load_path) = rb_get_expanded_load_path();
     if (load_path) {
 	long i;
 
diff --git a/internal.h b/internal.h
index e538de4..104539b 100644
--- a/internal.h
+++ b/internal.h
@@ -133,6 +133,7 @@ VALUE rb_iseq_clone(VALUE iseqval, VALUE newcbase);
 
 /* load.c */
 VALUE rb_get_load_path(void);
+VALUE rb_get_expanded_load_path(void);
 NORETURN(void rb_load_fail(VALUE, const char*));
 
 /* math.c */
diff --git a/load.c b/load.c
index 53dc58c..22d3532 100644
--- a/load.c
+++ b/load.c
@@ -74,7 +74,7 @@ load_path_getcwd(void)
     return cwd_str;
 }
 
-static VALUE
+VALUE
 rb_get_expanded_load_path(void)
 {
     rb_vm_t *vm = GET_VM();
-- 
1.7.10.2 (Apple Git-33)

0004-Fix-compatibility-of-require.patch

From f21777810c1dac501749218fe89d567dff39369b Mon Sep 17 00:00:00 2001
From: Hiroshi Shirosaki <[email protected]>
Date: Mon, 29 Oct 2012 23:22:01 +0900
Subject: [PATCH] Fix compatibility of require

* load.c (rb_construct_expanded_load_path): don't replace $LOAD_PATH.
  We just freeze contents of $LOAD_PATH. We add an argument for home
  directory expansion.

* load.c (rb_get_expanded_load_path): expand path which starts with ~
  always for compatibility.

* test/ruby/test_require.rb (TestRequire): add tests for require
  compatibility.
  - not modify load path contents
  - require when ENV['HOME'] is changed
---
 load.c                    |   26 +++++++++++++++++---------
 test/ruby/test_require.rb |   38 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+), 9 deletions(-)

diff --git a/load.c b/load.c
index 22d3532..9f707a7 100644
--- a/load.c
+++ b/load.c
@@ -34,7 +34,7 @@ rb_get_load_path(void)
 }
 
 static void
-rb_construct_expanded_load_path(int only_relative, int *has_relative)
+rb_construct_expanded_load_path(int only_tilde, int only_relative, int *has_relative)
 {
     rb_vm_t *vm = GET_VM();
     VALUE load_path = vm->load_path;
@@ -45,17 +45,21 @@ rb_construct_expanded_load_path(int only_relative, int *has_relative)
     ary = rb_ary_new2(RARRAY_LEN(load_path));
     for (i = 0; i < RARRAY_LEN(load_path); ++i) {
 	VALUE path, as_str, expanded_path;
+	char *as_cstr;
 	as_str = path = RARRAY_PTR(load_path)[i];
 	StringValue(as_str);
-	if (only_relative && rb_is_absolute_path(StringValuePtr(as_str))) {
+	as_cstr = StringValuePtr(as_str);
+	if (only_relative && rb_is_absolute_path(as_cstr)) {
 	    rb_ary_push(ary, RARRAY_PTR(expanded_load_path)[i]);
 	    continue;
 	}
-	if (!*has_relative && !rb_is_absolute_path(StringValuePtr(as_str)))
+	else if (only_tilde && (!as_cstr[0] || as_cstr[0] != '~')) {
+	    rb_ary_push(ary, RARRAY_PTR(expanded_load_path)[i]);
+	    continue;
+	}
+	if (!*has_relative && !rb_is_absolute_path(as_cstr))
 	    *has_relative = 1;
-	if (as_str != path)
-	    rb_ary_store(load_path, i, as_str);
-	rb_str_freeze(as_str);
+	rb_obj_freeze(path);
 	expanded_path = rb_file_expand_path_fast(as_str, Qnil);
 	rb_str_freeze(expanded_path);
 	rb_ary_push(ary, expanded_path);
@@ -82,17 +86,21 @@ rb_get_expanded_load_path(void)
     if (!rb_ary_shared_with_p(vm->load_path_snapshot, vm->load_path)) {
 	/* The load path was modified. Rebuild the expanded load path. */
 	int has_relative = 0;
-	rb_construct_expanded_load_path(0, &has_relative);
+	rb_construct_expanded_load_path(0, 0, &has_relative);
 	vm->load_path_cwd = has_relative ? load_path_getcwd() : 0;
     }
     else if (vm->load_path_cwd) {
 	VALUE cwd = load_path_getcwd();
+	int has_relative = 1;
 	if (!rb_str_equal(vm->load_path_cwd, cwd)) {
 	    /* Current working directory or filesystem encoding was changed.
 	       Expand relative load path again. */
-	    int has_relative = 1;
 	    vm->load_path_cwd = cwd;
-	    rb_construct_expanded_load_path(1, &has_relative);
+	    rb_construct_expanded_load_path(0, 1, &has_relative);
+	}
+	else {
+	    /* Expand only tilde (User HOME) */
+	    rb_construct_expanded_load_path(1, 0, &has_relative);
 	}
     }
     return vm->expanded_load_path;
diff --git a/test/ruby/test_require.rb b/test/ruby/test_require.rb
index 8c72571..ec7f2aa 100644
--- a/test/ruby/test_require.rb
+++ b/test/ruby/test_require.rb
@@ -457,4 +457,42 @@ class TestRequire < Test::Unit::TestCase
       }
     }
   end
+
+  def test_require_unmodified_lood_path
+    Dir.mktmpdir {|tmp|
+      Dir.chdir(tmp) {
+        open("foo.rb", "w") {}
+        assert_in_out_err([], <<-INPUT, %w(:ok), [])
+          $: << "#{tmp}"
+          a = 123
+          $: << a
+          begin
+            require "foo"
+          rescue TypeError
+          end
+          p :ok if $:.pop == a
+        INPUT
+      }
+    }
+  end
+
+  def test_require_changed_home
+    home = ENV['HOME']
+    Dir.mktmpdir {|tmp|
+      Dir.chdir(tmp) {
+        open("foo.rb", "w") {}
+        Dir.mkdir("a")
+        open(File.join("a", "bar.rb"), "w") {}
+        assert_in_out_err([], <<-INPUT, %w(:ok), [])
+          $: << '~'
+          ENV['HOME'] = "#{tmp}"
+          require "foo"
+          ENV['HOME'] = "#{tmp}/a"
+          p :ok if require "bar"
+        INPUT
+      }
+    }
+  ensure
+    ENV['HOME'] = home
+  end
 end
-- 
1.7.10.2 (Apple Git-33)