Activity
From 02/03/2021 to 02/09/2021
02/09/2021
-
11:46 PM Bug #17467: What makes strip() and lstrip() not stripping \u0000 from beginning of the string?
- I agree that this is a bug. The documentation for strip and lstrip specifies that the whitespace should be stripped, and defines whitespace to include null. I have submitted a pull request to fix it: https://github.com/ruby/ruby/pull/4164
-
11:42 PM Bug #17618: Exceptions in Fiber Scheduler causes a segv
- Just some observations in case it's useful. Implementing `unblock` in the scheduler and printing out the current thread shows that `unblock` runs on a dead thread:
```ruby
class Scheduler
def block blocker, timeout = nil
end
def ... -
11:12 PM Bug #17618 (Closed): Exceptions in Fiber Scheduler causes a segv
- If the fiber scheduler doesn't define an unblock function, Ruby will segv when threads are joined.
Here is an example program:
``` ruby
class Scheduler
def block blocker, timeout = nil
end
def fiber &block
fiber = ... -
- I've added a pull request that fixes the documentation and removes code that deals with a missing thread group: https://github.com/ruby/ruby/pull/4163
-
-
10:47 PM Revision 49d3830f (git): Fix documentation for Module#ruby2_keywords
- It returns nil, not self.
Fixes [Bug #17560] -
-
-
-
- This isn't a bug. Reline is calling `method` on an object, which by default is implemented by `Kernel#method`. By running `def method; end` in the top level scope, you are defining `Object#method`, which takes precedence over `Kernel#m...
-
09:23 PM Bug #17612: ractor crash with some simple codes
- While I don't have access to a build of ruby with debug symbols (and nmake DOES NOT like spaces in usernames), I have been able to reproduce a crash with Application Verifier enabled with heaps checks. It triggers an access violation see...
-
01:05 PM Bug #17612: ractor crash with some simple codes
- @ko1 Here is my error log when running your simpler code:
irb(main):036:0> 2.times.map{ Ractor.new{ loop{ rand.to_s } } }.each(&:take)
<internal:ractor>:267: warning: Ractor is experimental, and the behavior may change in future vers... -
02:31 AM Bug #17612 (Assigned): ractor crash with some simple codes
- Could you give me an error log?
```ruby
2.times.map{ Ractor.new{ loop{ rand.to_s } } }.each(&:take)
```
I can find an issue on the simpler example, but I also observed different error on the your original example.
I want to conf... -
- 09:16 PM Revision 189bf010 (git): Fix broken link
- The document should point to doc/syntax/comments.rdoc (with an "s") as there currently no `doc/syntax/comment.rdoc`
-
-
07:41 PM Feature #17615: Dump machine registers on RISC-V Linux
- Merged, thank you!
-
- Applied in changeset commit:git|66394aec9813e52751e2b099ee6a4814ace494f8.
----------
vm_dump.c: Dump machine registers on RISC-V Linux
Fixes [Bug #17615]
Co-Authored-By: xtkoba (Tee KOBAYASHI) <xtkoba+ruby@gmail.com> -
09:33 AM Feature #17615 (Closed): Dump machine registers on RISC-V Linux
- Dump every integer register that is either callee-saved or used for parameter passing on RISC-V Linux. An example output:
```
-- Machine register context ------------------------------------------------
sp: 0x0000004001801550 s0: 0... - 07:40 PM Revision bb10f81c (git): * 2021-02-10 [ci skip]
-
- Fixes [Bug #17615]
Co-Authored-By: xtkoba (Tee KOBAYASHI) <xtkoba+ruby@gmail.com> -
04:24 PM Bug #17617 (Closed): When a Ractor's incoming port is closed, Ractor.receive_if does not raise Ractor::ClosedError, but instead blocks indefinitely
- If Ractor#close_incoming is called on a Ractor, any pending Ractor.receive call will raise Ractor::ClosedError. However, any pending Ractor.receive_if call will not; instead, it never returns.
```
r1 = Ractor.new do
Ractor.receive... - 01:54 PM Revision ccba4665 (git): * 2021-02-09 [ci skip]
-
01:53 PM Revision 09c681ab (git): Merge RubyGems 3.2.9 and Bundler 2.2.9 (#4158)
- 01:22 PM Revision e7a831de (git): Improve the wording surrounding `Proc#isolate`
- This change moves the statement that `Proc#isolate` isn't yet exposed for Ruby users to the first time the method is mentioned. This is so that readers don't waste time trying to look it up in Ruby docs.
There was also unnecessary dupli... -
- Currently, `vm_dump.c` assumes that the availability of `backtrace` implies the existence of `execinfo.h`, but it is not the case with non-GNU libc (e.g. musl or Bionic) Linux systems where libunwind is installed.
A patch is attached ... -
08:49 AM Revision 6bb103f0 (git): [ruby/console] Updated depend
-
-
-
-
-
- 06:39 AM Revision 5f69a7f6 (git): Expose scheduler as public interface & bug fixes. (#3945)
- * Rename `rb_scheduler` to `rb_fiber_scheduler`.
* Use public interface if available.
* Use `rb_check_funcall` where possible.
* Don't use `unblock` unless the fiber was non-blocking. -
05:48 AM Revision 3c593f28 (git): Fixed default coroutine selection for musl.
-
04:30 AM Feature #17614 (Closed): Improve performance Float#negative? and Float#positive?
- Applied in changeset commit:git|fad7908a5de4ab08367914d53780ff6518d5f552.
----------
Improve performance Float#positive? and Float#negative? [Feature #17614] (#4160) -
- Improve performance `Float#negative?` and `Float#positive?`(write in Ruby)
benchmark:
```yml
prelude: |
flo = 4.2
benchmark:
negative?: |
flo.negative?
positive?: |
flo.positive?
loop_count: 20000000
```
r... -
-
- The list is reused when an exception raised again after retrying
in the rescue procedure. -
02/08/2021
-
-
- How do I assign? It looks like this should go to @ko1 (eval) or @nobu (mingw)
-
- Still active as of Ruby 3.0.0p0 (x64-mingw32). I've done some more digging, and it appears to be freeing the Thread_Mutex in rb_call_inits (inits.c), aka the Init_Thread_Mutex function in thread.c.
-
- This patch frees catch tables on iseqs that don't use the catch tables. It also eliminates `nop` instructions from lambdas that don't need them.
Before this patch, lambdas have a "prelude nop" that is used for catch table entries:
... -
-
- 03:07 PM Revision 15e933f2 (git): * 2021-02-09 [ci skip]
-
-
-
- * arguments are optional
* `path` may be passed as a keyword argument
* `nil` or `"-"` path means stdout -
- This reverts commit 478716f49a19cdd86f629c6a0673c1ff53630c96
partially -- "Removed unnecessary parentheses" part -- because of
a lambda, see 9e25eb308d4fae9a10e120c2b4601916cc38336c. -
10:54 AM Bug #17482: Ractor.receive doesnt accept Ctrl+C under some condition
- ruby_3_0 9aba46d8d80473594e567dff1652626e7b2a77b0 merged revision(s) 6f727853cee41195b67ee5d793c1ac23fe1a6ae0,b2674c1fd725b43adf51af7935c780359d70c2a4.
-
- only main thread can modify vm->ubf_async_safe
vm->ubf_async_safe is VM global resource and only main thread
can manipulate it.
[Bug #17482]
---
thread.c | 8 +++++---
1 file changed, 5 i... -
09:56 AM Bug #17599: [BUG] try to mark T_NONE object when object allocation tracing is enabled
- @wanabe I just tested your patch and it seems to work. Thank you!
-
- I found these codes will crash because of **Ractor** :
``` ruby
e = (1..1000).to_a
arr = [e,e,e,e]
arr.map { |sub_arr|
Ractor.new(sub_arr) do |sub_arr|
sub_arr.map {
Digest::MD5.hexdigest(ra... -
- Could you show mkmf.log file for a failed extension?
-
04:06 AM Feature #17601 (Closed): lib/benchmark: adding `Benchmark::Tms#to_h`
- Applied in changeset commit:git|3a7ff66abc0c389851ab3c5cd0ac53ebe0647eec.
----------
[ruby/benchmark] Adds `Tms#to_h`
[Feature #17601] -
- [Feature #17601]
- 02:38 AM Revision 9328112b (git): * 2021-02-08 [ci skip]
-
-
-
02/07/2021
-
11:41 PM Bug #17486: Build fails on darwin due to libtool being removed
- I'm running into this exact same issue, it fails on both Catalina and Big Sur.
```
$ ASDF_RUBY_BUILD_VERSION=master RUBY_CONFIGURE_OPTS="--with-openssl-dir=$(brew --prefix openssl) --with-readline-dir=$(brew --prefix readline)" asdf ins... -
-
-
-
- * Adjusted indentation in license section, and used "(c)"
* Commented out invalid syntax to enable highlighting
* Removed unnecessary parentheses - 12:04 PM Revision 6e8e7005 (git): * 2021-02-07 [ci skip]
-
12:04 PM Revision 77700bf0 (git): Backport lib/reline, and lib/irb for 3.0.1 2nd (#4157)
- * [ruby/irb] Stub a screen size for tests
https://github.com/ruby/irb/commit/6663057083
* [ruby/irb] Support GitHub Actions
https://github.com/ruby/irb/commit/8e9e6c4037
* [ruby/irb] Stub a screen size for test_context
http://ci.rvm... -
10:03 AM Bug #17587: Segmentation fault with ractors and unix signals
- The current behavior of stacking appears to be as expected.
The registered signal handler is called from the main thread. https://git.ruby-lang.org/ruby.git/tree/thread.c?id=947d93b715436b13eefa39f87737bdad3c1f870a#n2430
`Ractor.yiel... -
- An opaque `struct rb_execarg` has been added 9 years ago, and the latest change on functions deal with it was 3 years ago.
Also, it has been 3 years since `rb_grantpt` was added.
These seem stable enough to expose now, but are still in... -
02/06/2021
-
- https://github.com/ruby/reline/commit/b26c7d60c8
-
- https://github.com/ruby/irb/commit/4c87035b7c
-
- https://github.com/ruby/reline/commit/e1d9240ada
-
- https://github.com/ruby/reline/commit/ebaf37255f
-
- This closes ruby/reline#253.
https://github.com/ruby/reline/commit/f131f86d71 -
- https://github.com/ruby/reline/commit/ba06e4c480
-
- https://github.com/ruby/reline/commit/92d314f514
-
- https://github.com/ruby/reline/commit/f197139b4a
-
- On Unix-like OSes, logs prior to the screen are not editable. When the code
is higher than the screen, the code is only shown on the screen until input
is finished, but when it is finished, all lines are outputted.
https://github.com/ru... -
- https://github.com/ruby/reline/commit/6877a7e3f5
-
- https://github.com/ruby/reline/commit/123ea51166
-
- https://github.com/ruby/reline/commit/25af4bb64b
-
- Because it's too slow.
The rendering time in IRB has been reduced as follows:
start = Time.now
def each_top_level_statement
initialize_input
catch(:TERM_INPUT) do
loop do
begin
prompt
unle... - 06:03 PM Revision 132e01a1 (git): .gdbinit: Fix indentation in print_id [ci skip]
- 06:02 PM Revision bfc7bb5d (git): * 2021-02-07 [ci skip]
- 06:02 PM Revision 34c422b6 (git): Fix grammatical error
-
- https://github.com/ruby/irb/commit/b444573aa2
-
- https://github.com/ruby/irb/commit/20f1ca23e9
-
- Shouldn't the call to `gc_event_hook()` come after `newobj_fill()`?
Like this:
```patch
diff --git a/gc.c b/gc.c
index 30badf975f..dc90e6177f 100644
--- a/gc.c
+++ b/gc.c
@@ -2254,7 +2254,6 @@ newobj_slowpath(VALUE klass, VALUE ... -
10:47 AM Feature #15504: Freeze all Range objects
- > if I want to check was called refined method or core?
It actually might be a good feature proposal for Ruby. Because currently, you can tell whether the method is defined by this class, by its parent, by included module, by singleto... -
12:29 AM Feature #15504: Freeze all Range objects
- zverok (Victor Shepelev) wrote in #note-15:
> @AlexWayfer
> ...
Thank you, I agree, it's better. But… if I want to check was called refined method or core? Right now I'm doing it via `have_received` once or never (depending on Ruby versi... -
08:27 AM Bug #16651: Extensions Do Not Compile on Mingw64 with mingw32-make
- Sorry, missed your comment. The problem is /C/ is *wrong*.
This is still an issue with Ruby 3.0.0 and causes building extensions to fail.
Install msys2, install mingw64. Then try to build any extension. You'll get this error message:
... -
08:23 AM Revision e1e61e25 (git): Improve performance of Array#- when it is called with empty array
- This change make Array#- return a copy of the receiver when
the other array is empty. -
06:28 AM Bug #16455: coroutine ucontext uses deprecated POSIX getcontext/swapcontext/makecontext, absent in musl and uclibc
- I still the get same issue with 3.0.0.
Ruby cannot blindly try to use ucontext without checking for getcontext swapcontext makecontext presence.
https://github.com/ruby/ruby/pull/3567 does work. It only needs to be refreshed, removin... -
- 04:14 AM Revision 575c77dd (git): * 2021-02-06 [ci skip]
-
-
01:14 AM Misc #17535: DevelopersMeeting20210216Japan
- * [Feature #17608] Compact and sum in one step (sawa)
* Let `Array#sum` ignore `nil` values, or introduce a method that does that.
* Many actual use cases of `Array#sum` are followed by `compact`.
* If we need to take care of...
02/05/2021
-
08:05 PM Feature #17608: Compact and sum in one step
- I closed it.
It seems clear there is no need for a new method or keyword argument, when `a.sum{_1 || 0}` already works well, is clear and concise.
In general it's a good idea to think about what to do with missing data anyway, ignoring... -
-
04:17 PM Feature #17608: Compact and sum in one step
- Then use `sum(&:to_f)`...
I agree, we should close this request. -
- Hanmac (Hans Mackowiak) wrote in #note-6:
> [Y]ou can use [the fact] that `nil.to_i` returns 0
> ...
That won't work.
```ruby
[1.2, nil, 3.6].sum{_1 || 0} # => 4.8
[1.2, nil, 3.6].sum(&:to_i) # => 4
```
-
01:27 PM Feature #17608: Compact and sum in one step
- you can use that `nil.to_i` returns 0
`a.sum(&:to_i) #=> 6` -
- b2674c1fd725b43adf51af7935c780359d70c2a4 should be backported too.
-
- other ractors should not have a unblock thread.
This patch fixes 6f727853cee41195b67ee5d793c1ac23fe1a6ae0. -
-
-
03:29 AM Revision abf71864 (git): [ruby/io-console] Mark all tests that use run_pty as pend on JRuby
- JRuby's PTY.spawn does not produce a process with its own
controlling terminal, which is necessary for testing these raw
escape sequences. This commit marks those tests as pending.
The functionality tested appears to work at a command l... -
- This test runs with test/unit now, which defines omit instead of
skip.
https://github.com/ruby/io-console/commit/bd731d0b8d -
- The subprocess script here works fine at a command line, but when
run as a pty subprocess during the tests the master side hangs
waiting for output.
https://github.com/ruby/io-console/commit/4a21610ece -
- https://github.com/ruby/io-console/commit/74567376c4
-
- Now io/console is loaded from test-unit indirectly, test-unit
requires power_assert which requires io/console.
https://github.com/ruby/io-console/commit/8817d07951
https://github.com/ruby/io-console/commit/639cce89de -
- At least check_dependencies.yml has not failed here yet.
02/04/2021
-
- `a.sum{_1 || 0}` seems more than good enough.
I don't think we want composite methods unless there is a significant performance advantage and it's more expressive. -
- Many use cases of `Array#sum` are preceded with the `compact` method or are followed by a block to ensure the value is addable.
```ruby
a = [1, nil, 2, 3]
a.sum # !> TypeError
a.compact.sum # => 6
a.sum{_1 || 0} # => 6
```
... -
- 05:49 PM Revision 365326a0 (git): * 2021-02-05 [ci skip]
-
- `rb_define_const` can add objects as "mark objects". This is to make
code like this work:
https://github.com/ruby/ruby/blob/33d6e92e0c6eaf1308ce7108e653c53bb5fb106c/ext/etc/etc.c#L1201
```
rb_define_const(rb_cStruct, "Passwd", s... -
12:12 PM Revision b79d4434 (git): [ruby/irb] Add info.rb to gemspec
- https://github.com/ruby/irb/commit/adbba19adf
-
- Pull Request: https://github.com/ruby/ruby/pull/4119
### Context
While profiling our application allocations, I noticed `load_from_binary` would allocate a string for each method:
```
305.68 kB 7642 "initialize"
... -
- Thanks! I've committed your patch. I'll check the RISC-V CI later: http://rubyci.s3.amazonaws.com/debian-riscv64/ruby-master/recent.html
-
- Applied in changeset commit:git|7d22060c14895bbca3ff962e5456384095b66738.
----------
addr2line.c: DW_LNS_fixed_advance_pc takes a single uhalf operand
Fixes [Bug #17609]
Co-Authored-By: xtkoba (Tee KOBAYASHI) <xtkoba+ruby@gmail.com> -
- According to the standards, DW_LNS_fixed_advance_pc takes a single uhalf operand. A patch is attached to conform to them.
With this change, we can safely revert the commit 6985bfb72c8d7e7d94f1aad1f0902b602bbfa6da (vm_dump.c: Do not sh... -
- This reverts commit 6985bfb72c8d7e7d94f1aad1f0902b602bbfa6da.
Now it should work on riscv by the previous commit. [Bug #17609] -
- Fixes [Bug #17609]
Co-Authored-By: xtkoba (Tee KOBAYASHI) <xtkoba+ruby@gmail.com> -
- 04:29 AM Revision 46b5b51c (git): [DOC] Fix typo in ext/socket/socket.c [ci skip]
- 04:18 AM Revision 4e2e1d60 (git): [ruby/bigdecimal] Fix uint64 conversion
- Stop using logarithm to compute the number of components.
Instead, use the theoretical maximum number of components for buffer,
and count up the actual number of components during conversion.
https://github.com/ruby/bigdecimal/commit/90... - 04:18 AM Revision 868d66e0 (git): [ruby/bigdecimal] Stop using rmpd and RMPD prefixes and DBLE_FIG
- https://github.com/ruby/bigdecimal/commit/7479923fdb
-
03:34 AM Revision b0d6563f (git): rb_clear_method_cache_by_class: no longer exist
- It is questionable whether we can change a public API but this function
does not exist today. No way.
02/03/2021
-
- I think this is breaking something
This reverts commit 1be84e53d76cff30ae371f0b397336dee934499d. - 05:35 PM Revision 6b481408 (git): * 2021-02-04 [ci skip]
-
- The caller should be responsible for holding a pinned reference (if they
need that) -
04:13 PM Bug #17607: ChildProcess vs RUBY_PIPE_NONBLOCK_DEFAULT
- vo.x (Vit Ondruch) wrote:
> Unfortunately, other test case started to fail then. I am not really sure what might be the right fix.
Starting from scratch again, this might be fix after all, because I can't see any issues now. Ruby as ... -
- I am investigating why ChildProcess test suite fails running against Ruby 3.0 [1]. The current failure is:
~~~
1) ChildProcess can write to stdin interactively if duplex = true
Failure/Error: raise msg
RuntimeErr... -
-
02:36 PM Feature #16975: Warn when value from environment is being ignored by `Dir.tmpdir`
- Actually I don't think I have permissions to close the ticket, or at least I don't know how.
-
- I think @nobu fixed this at https://github.com/ruby/tmpdir/pull/7, so I'm closing this. Thanks!
I'd like to ask why issues were disabled at https://github.com/ruby/tmpdir. -
- Chaining aliased methods increases searching cost linearly.
- 09:25 AM Revision 8df1881c (git): [ruby/bigdecimal] Fix the maximum length of float number
- This change is for preventing the false-positive alert by CoverityScan.
See CID-1471770 for the detail.
https://github.com/ruby/bigdecimal/commit/4d5b97125b -
-
-
- "alias" type method entries can chain another aliased method
so that machine stack can be overflow on nested alias chain.
http://ci.rvm.jp/results/trunk-repeat20@phosphorus-docker/3344209
This patch fix this issue by use goto instead of... -
- @xtkoba Awesome!
@vo.x Can you check the patch on your Fedora machines? -
- "A segfault in the segfault handler" occurs also on my x86_64-linux and aarch64-linux when `gcc-10.2.0 -gdwarf-5` is used. A backtrace with GDB:
```
(gdb) run -e 'Process.kill :SEGV, $$'
Starting program: /var/tmp/ruby.build/ruby-3.0.... -
05:38 AM Revision f600226f (git): Fix failures with EditLine
- On macOS with EditLine:
```
$ ruby -r readline -e 'Readline.readline("> "); p Readline::VERSION' < /dev/null
"EditLine wrapper"
```
On Linux with GNU readline:
```
$ ruby -r readline -e 'Readline.readline("> "); p Readline::VERSION' < /... -
05:03 AM Bug #17486: Build fails on darwin due to libtool being removed
- Adding a data point to this. I've also tried reinstalling the CLTs on Catalina, but the build only succeeded with the reverted commit. Big Sur doesn't support my iMac, so I'm stuck with Catalina for good.
-
- Applied in changeset commit:git|6f727853cee41195b67ee5d793c1ac23fe1a6ae0.
----------
only main thread can modify vm->ubf_async_safe
vm->ubf_async_safe is VM global resource and only main thread
can manipulate it.
[Bug #17482] -
-
- これがあると何がどう嬉しいのでしょうか。
少なくともインデントルールをrubyに導入するのはかなり困難だと思います。慣習的にも実装的にも。