[#28395] [Bug #2830] Some methods raise ArgumentError instead of TypeError — Marc-Andre Lafortune <redmine@...>
Bug #2830: Some methods raise ArgumentError instead of TypeError
[#28405] [Feature #2832] Vector#each and Enumerable — Marc-Andre Lafortune <redmine@...>
Feature #2832: Vector#each and Enumerable
[#28452] Watched issues on redmine — Caleb Clausen <vikkous@...>
Is there a page on redmine that will show me the list of issues that
[#28482] Question on scoped constant resolution Class vs Module — Peter McLain <peter.mclain@...>
I asked this on ruby-talk, but didn't get anywhere. Someone suggested
[#28505] [Bug #2838] Ruby 1.8.7 (2009-06-12 patchlevel 174) strange round behaviour — P K <redmine@...>
Bug #2838: Ruby 1.8.7 (2009-06-12 patchlevel 174) strange round behaviour
[#28552] [Bug #2945] Regexp#=== is failed by an exception when the exception is occurred in method_missing — Kenta Murata <redmine@...>
Bug #2945: Regexp#=== is failed by an exception when the exception is occurred in method_missing
Hi,
Hi,
Hi,
Hi,
[#28561] Ruby::DL vs Ruby::FFI — Aston <blackapache512-ticket@...>
Ruby.DL and FFI libraries are great for programmers like me who are not internet programmers, but are more interested in scientific and number processing etc.
On Mon, Mar 8, 2010 at 4:56 PM, Aston <blackapache512-ticket@yahoo.com> wrote:
[#28576] "rake not found" error on a rubygems test — Yusuke ENDOH <mame@...>
Hi Eric Hodel,
[#28583] build failure on 26861 using msys/mingw — Jon <jon.forums@...>
Can anyone replicate? I've recently updated both binutils and the mingw runtime so this may very well be my configuration.
[#28602] [Bug #2952] Time.strftime format %N — Russell Penney <redmine@...>
Bug #2952: Time.strftime format %N
[#28643] [Bug #2957] IO.print emits field separator after each object, rather than between — Daniel Kelley <redmine@...>
Bug #2957: IO.print emits field separator after each object, rather than between
[#28665] [ANN] 1.9.2 release plan — Yusuke ENDOH <mame@...>
Hi,
[#28686] trunk (26947) build fail with msys/mingw/vista — Jon <jon.forums@...>
I get the following build failure when msysgit's "c:\git\cmd" dir is on PATH.
On Tue, Mar 16, 2010 at 5:50 AM, Jon <jon.forums@gmail.com> wrote:
[#28712] When a trace hook raises an exception, should it terminate the program? — Rocky Bernstein <rockyb@...>
In Ruby 1.8 and the Ruby 1.9 trunk when running a trace hook that raises an
On Wed, Mar 17, 2010 at 5:42 AM, Rocky Bernstein <rockyb@rubyforge.org> wrote:
Let me clarify a bit because I think some of the facts (some by me) may have
On Thu, Mar 18, 2010 at 5:52 AM, Rocky Bernstein <rockyb@rubyforge.org> wrote:
[#28724] [Feature:trunk] Array#repeated_(permutation|combination) — "KISHIMOTO, Makoto" <ksmakoto@...4u.or.jp>
New methods Array#repeated_(permutation|combination).
[#28735] [Bug #2982] Ruby tries to link with both openssl and readline — Lucas Nussbaum <redmine@...>
Bug #2982: Ruby tries to link with both openssl and readline
Issue #2982 has been updated by caleb clausen.
Hi,
[#28783] [Feature #2065] An ancestors iterator — Simon Chiang <redmine@...>
Issue #2065 has been updated by Simon Chiang.
Hi,
[#28837] [Bug #2993] Module#instance_methods' flag seems to be ignored in singleton classes — Xavier Noria <redmine@...>
Bug #2993: Module#instance_methods' flag seems to be ignored in singleton classes
[#28859] st.c: pool allocator for tables and entries — Eric Wong <normalperson@...>
Hi all,
[#28865] Can DRb be used across a fork() — Chris Schlaeger <cschlaeger@...>
I'm trying to use DRb to communicate between a parent and child
[#28871] WeakRef extending Delegator is a bug waiting to happen? — Charles Oliver Nutter <headius@...>
Hopefully this doesn't contradict my other email too much :)
[#28902] [Bug #2998] gets fails in mingw — Roger Pack <redmine@...>
Bug #2998: gets fails in mingw
[#28907] [Bug #3000] Open SSL Segfaults — Christian Höltje <redmine@...>
Bug #3000: Open SSL Segfaults
Issue #3000 has been updated by Hiroshi NAKAMURA.
Hi,
Hi,
Hi,
[#28924] [Bug #3005] Ruby core dump - [BUG] rb_sys_fail() - errno == 0 — Sebastian YEPES <redmine@...>
Bug #3005: Ruby core dump - [BUG] rb_sys_fail() - errno == 0
[#28954] [Feature #3010] slow require gems in ruby 1.9.1 — Miao Jiang <redmine@...>
Feature #3010: slow require gems in ruby 1.9.1
[#29019] [Bug #3015] NetBSD vs test/dl — Michael Graff <redmine@...>
Bug #3015: NetBSD vs test/dl
On Fri, Mar 26, 2010 at 11:49:59AM +0900, Michael Graff wrote:
[#29031] [Feature #1395](Open) Steppable Kernel::eval — Yusuke Endoh <redmine@...>
Issue #1395 has been updated by Yusuke Endoh.
[#29045] [Feature #3021] Array#product should accept a block. — Marc-Andre Lafortune <redmine@...>
Feature #3021: Array#product should accept a block.
[#29092] merged psych to trunk — Aaron Patterson <aaron@...>
Hey everyone,
[#29118] [Bug #3051] psych is too osx-specifc — Michael Graff <redmine@...>
Bug #3051: psych is too osx-specifc
[#29128] [Bug #3052] DRb::start_service fails to detect used port — Chris Schlaeger <redmine@...>
Bug #3052: DRb::start_service fails to detect used port
Issue #3052 has been updated by Yusuke Endoh.
[#29131] [trunk:bug] Many rubygems tests fail with psych tests. — Tanaka Akira <akr@...>
Many rubygems tests fail with psych tests.
(2010/03/30 17:55), Tanaka Akira wrote:
On Tue, Mar 30, 2010 at 07:13:32PM +0900, NARUSE, Yui wrote:
[#29161] [Bug #3058] Inconsistent eol conversion of IO#read on Windows — Heesob Park <redmine@...>
Bug #3058: Inconsistent eol conversion of IO#read on Windows
[#29167] [Feature #3067] complex.c : Question: why Complex#~ is disabled? It's in the doc — Benoit Daloze <redmine@...>
Feature #3067: complex.c : Question: why Complex#~ is disabled? It's in the doc
[#29179] [Bug #3071] Convert rubygems and rdoc to use psych — Aaron Patterson <redmine@...>
Bug #3071: Convert rubygems and rdoc to use psych
Doesn't this mean the the RubyGems codevase would now be forked
Issue #3071 has been updated by Nobuyoshi Nakada.
[#29186] [Bug #3072] Classes Inheriting from Data — Run Paint Run Run <redmine@...>
Bug #3072: Classes Inheriting from Data
[ruby-core:28859] st.c: pool allocator for tables and entries
Hi all,
I started playing around with pool allocator for Ruby 1.9 for hashes. I
cooked this up for an app that generates lots of Hashes. After getting
some performance improvements from tweaking constants inside gc.c for
the app and using bsdmalloc[1], I still got a ~10% runtime reduction
with this small change (for one particular app).
This assumes st.c is only called inside GVL, and GVL == GIL which is
true these days in 1.9.1 and trunk.
The only other downside is that memory used for each type of struct
cannot be used anything else, nor released back to the OS (though some
malloc implementations don't do this anyways).
I've sized both of the pools conservatively (8K) for now.
Further testing and benchmarks results on different apps and comments
would be greatly appreciated.
This patch should apply cleanly to trunk, and works with 1.9.1, too.
[1] I used bsdmalloc from here: git://github.com/ice799/bsdmalloc
---
st.c | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++++----------
1 files changed, 54 insertions(+), 10 deletions(-)
diff --git a/st.c b/st.c
index ec518e9..cbc57d8 100644
--- a/st.c
+++ b/st.c
@@ -38,6 +38,50 @@ struct st_table_entry {
*
*/
+#if __GNUC__ >= 3
+#define UNLIKELY(x) (__builtin_expect((x), 0))
+#else /* __GNUC__ >= 3 */
+#define UNLIKELY(x) (x)
+#endif /* __GNUC__ >= 3 */
+
+#define DEFINE_ALLOC_POOL(name,type,size) \
+struct name##_alloc_ptr { \
+ union { \
+ type data; \
+ void *next; \
+ } as; \
+}; \
+static struct name##_alloc_ptr *name##_head; \
+static void more_##name(void) \
+{ \
+ long i; \
+ struct name##_alloc_ptr *cur, *tmp; \
+ tmp = ruby_xmalloc((size / sizeof(type)) * sizeof(type)); \
+ name##_head = tmp; \
+ for (i = size / sizeof(type); --i >= 0; ) { \
+ cur = tmp++; \
+ cur->as.next = tmp; \
+ } \
+ cur->as.next = NULL; \
+} \
+static type *alloc_##name(void) \
+{ \
+ struct name##_alloc_ptr *tmp; \
+ if (UNLIKELY(!name##_head)) more_##name(); \
+ tmp = name##_head; \
+ name##_head = name##_head->as.next; \
+ return (type *)tmp; \
+} \
+static void release_##name(void *ptr) \
+{ \
+ struct name##_alloc_ptr *tmp = ptr; \
+ tmp->as.next = name##_head ? name##_head : NULL; \
+ name##_head = tmp; \
+}
+
+DEFINE_ALLOC_POOL(entry, struct st_table_entry, 0x2000);
+DEFINE_ALLOC_POOL(table, struct st_table, 0x2000);
+
static const struct st_hash_type type_numhash = {
st_numcmp,
st_numhash,
@@ -183,7 +227,7 @@ st_init_table_with_size(const struct st_hash_type *type, st_index_t size)
size = new_size(size); /* round up to prime number */
- tbl = alloc(st_table);
+ tbl = alloc_table();
tbl->type = type;
tbl->num_entries = 0;
tbl->entries_packed = type == &type_numhash && size/2 <= MAX_PACKED_NUMHASH;
@@ -253,7 +297,7 @@ st_clear(st_table *table)
table->bins[i] = 0;
while (ptr != 0) {
next = ptr->next;
- free(ptr);
+ release_entry(ptr);
ptr = next;
}
}
@@ -267,7 +311,7 @@ st_free_table(st_table *table)
{
st_clear(table);
free(table->bins);
- free(table);
+ release_table(table);
}
size_t
@@ -394,7 +438,7 @@ do {\
bin_pos = hash_val % table->num_bins;\
}\
\
- entry = alloc(st_table_entry);\
+ entry = alloc_entry(); \
\
entry->hash = hash_val;\
entry->key = key;\
@@ -562,7 +606,7 @@ st_copy(st_table *old_table)
st_index_t num_bins = old_table->num_bins;
st_index_t hash_val;
- new_table = alloc(st_table);
+ new_table = alloc_table();
if (new_table == 0) {
return 0;
}
@@ -572,7 +616,7 @@ st_copy(st_table *old_table)
Calloc((unsigned)num_bins, sizeof(st_table_entry*));
if (new_table->bins == 0) {
- free(new_table);
+ release_table(new_table);
return 0;
}
@@ -585,7 +629,7 @@ st_copy(st_table *old_table)
prev = 0;
tail = &new_table->head;
do {
- entry = alloc(st_table_entry);
+ entry = alloc_entry();
if (entry == 0) {
st_free_table(new_table);
return 0;
@@ -650,7 +694,7 @@ st_delete(register st_table *table, register st_data_t *key, st_data_t *value)
REMOVE_ENTRY(table, ptr);
if (value != 0) *value = ptr->record;
*key = ptr->key;
- free(ptr);
+ release_entry(ptr);
return 1;
}
}
@@ -722,7 +766,7 @@ st_cleanup_safe(st_table *table, st_data_t never)
if (ptr->key == never) {
tmp = ptr;
*last = ptr = ptr->next;
- free(tmp);
+ release_entry(tmp);
}
else {
ptr = *(last = &ptr->next);
@@ -798,7 +842,7 @@ st_foreach(st_table *table, int (*func)(ANYARGS), st_data_t arg)
tmp = ptr->fore;
*last = ptr->next;
REMOVE_ENTRY(table, ptr);
- free(ptr);
+ release_entry(ptr);
if (ptr == tmp) return 0;
ptr = tmp;
break;
--
Eric Wong