Activity
From 08/21/2021 to 08/27/2021
08/27/2021
-
02:13 PM Revision ed31bdfe (git): Fix memory leak in Variable Width Allocation
- Force recycled objects could create a freelist for the page. At the
start of sweeping we should append to the freelist to avoid
permanently losing the slots on the freelist. -
10:22 AM Feature #18137 (Open): A new method to check Proc is isolated or not
- I want to check a Proc is isolated or not, like `Proc#isolated?`.
Proc objects are passed to libraries very often. For example, Rack web application is a callable (`respond_to(:call)`) object, and it may be a Proc.
When the library w... -
09:28 AM Misc #18122: DevelopersMeeting20210916Japan
- * [Feature #18136] `take_while_after` (or `take_upto`)
* I have proposed it once a long time ago, but those examples were convoluted; now I came with clearer ones and explanations of the usefulness.
* I am ready to work on the PR, ... -
- Sorry, I already tried that once (#16441) but I failed to produce the persuasive example.
So I am back with a couple of them, much simpler and clear than my initial.
**The proposal itself:** Have `take_while_after` which behaves like... -
03:43 AM Revision 3562c7ba (git): Update the list of replacing objects under missing
-
-
-
02:15 AM Bug #18134: Memory leak with master
- The leak first appears after [`0ef2923c2b` Avoid rehashing in Hash#replace/dup/initialize_copy \[Bug #16996\]](https://github.com/ruby/ruby/commit/0ef2923c2b9afb76b3a1cebc56ebabc71c9a2ba8).
Log I kept while bisecting:
```
FAIL b990... -
-
-
-
- 12:16 AM Revision db4bfdfb (git): * 2021-08-27 [ci skip]
-
12:16 AM Revision 19ab24a0 (git): [DOC] Fix a typo in Dir.fnmatch
08/26/2021
-
10:41 PM Bug #9659 (Closed): crash in FIPS mode after unchecked algo->init_func failure
- The crash in FIPS mode was fixed 7 years ago. The reason this issue was reopened 7 years ago was because of digest/md5 using the low-level OpenSSL API instead of the EVP API. The digest extension removed the OpenSSL engine in commit:2e60...
-
07:18 PM Feature #18135 (Open): Introduce Enumerable#detect_only
- It can be important to get the only element for which a block returns true, and to assert that this is the only element as such. For example, this can be a very helpful sanity check when one is working with data that's outside of their c...
-
- While working on code to test Puma’s performance, I noticed a memory leak using WSL2 Ubuntu 20.04 and Ruby master. The leak seemed proportional to the number of requests made, and independent of the response size, either headers or body...
-
10:47 AM Revision ef10e8a1 (git): Use patch installed by msys2/setup-msys2
-
-
- Because patch does not installed on windows-2022.
https://github.com/actions/virtual-environments/issues/3949
> Only pure MSYS2 is installed. No additional packages -
- https://github.com/ruby/ruby/pull/4775/checks?check_run_id=3419490832#step:13:38
-
-
-
- https://github.blog/changelog/2021-08-23-github-actions-windows-server-2022-with-visual-studio-2022-is-now-available-on-github-hosted-runners-public-beta/
-
09:49 AM Revision 9c058270 (git): Fix example fiber scheduler reg. writable events
- There were two issues:
1. When an IO object is waiting for writablility only (as in test_tcp_accept) the selected hash is empty.
Therefore selected[fiber] returns nil but needs to default to 0 in order to be or'ed with IO::WRITABLE.
... - 04:29 AM Revision a2ad0cb7 (git): Make Hash#each family usable in Ractor
- We don't need to increment/decrement iteration level for frozen Hash
because frozen Hash can't be modified. We can assume that nobody
changes the target Hash while calling #each family.
How to reproduce:
a = {}
100.times do |i|... - 01:24 AM Revision 27410eae (git): * 2021-08-26 [ci skip]
-
01:23 AM Revision f4b88959 (git): Clean up lambda output from exyacc.rb
- The `lambda_body` grammar rule has a `"}"`, which is throwing off the `exyacc.rb` regular expressions. This changes the regular expression to account for `"}"` as well, which makes the output of `ruby sample/exyacc.rb < parse.y` change b...
08/25/2021
-
03:30 PM Bug #18133: LTO: TestGCCompact#test_ast_compacts segfaults on i686
- It actually crashes on c9s:
https://kojihub.stream.rdu2.redhat.com/koji/taskinfo?taskID=636003
https://kojihub.stream.rdu2.redhat.com/kojifiles/work/tasks/6017/636017/build.log -
- The backtrace looks similar to #18119 which is triggered in Ractor.
-
- I observe following segfault running the test suite on i686 on RHEL9:
~~~
$ gdb --args ./miniruby -I./lib -I. -I.ext/common ./tool/runruby.rb --extout=.ext -- --disable-gems ./test/runner.rb --excludes-dir=./test/excludes -v
...... -
- Adding the core Async gem to stdlib requires to add 4 more dependent gems. Async is a big library with a lot of supplemental gems. It makes things more complicated if some gems are in stdlib and some are not. Also Async is made to be usa...
-
-
02:13 PM Bug #17052: Ruby with LTO enabled has issues with SIGSEGV handler
- The situation `ruby-addr2line-abbrev_table-r1.patch` assumes is not likely to happen in the real world. Let's leave it unapplied for now. Thank you.
-
- xtkoba (Tee KOBAYASHI) wrote in #note-31:
> I might have been misleading. The patch in #note-9 is necessary for correctly reading `DW_FORM_ref_addr` values in `.debug_info` sections, even if those values are kept uninterpreted by the pa... -
- This commits implements size classes in the GC for the Variable Width
Allocation feature. Unless `USE_RVARGC` compile flag is set, only a
single size class is created, maintaining current behaviour. See the
redmine ticket for more detail... -
- This commit removes T_PAYLOAD since the new VWA implementation no longer
requires T_PAYLOAD types.
Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org> -
09:47 AM Revision 01be881f (git): Use `empty` instead of NOOP in mk file to make it safer
- This would avoid unintentional use of the `NOOP` environment variable
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org> -
- ./spec/ruby/optional/capi/ext/array_spec.c can match with spec/%/ if
using GNU Make under version 3.81. make command installed on macOS is
3.81, so ruby can't be built with default make on macOS with
--enable-shared option since https:... -
-
- This library does not consider multi-threading at all.
Callers must consider it instead.
AFAIK, VM list has a problem, and others may. -
06:02 AM Revision 384c0f57 (git): Added bundle install before bundle exec on rdoc sync
-
02:13 AM Revision b44741f5 (git): tool/test-bundled-gems.rb: Use the bundled RBS code to test TypeProf
- Formerly, TypeProf was tested with the latest RBS code during
`make test-bundled-gems`. However, when a new version of rbs is
released, and if it is incompatible with TypeProf,
`make test-bundled-gems` starts failing, which was annoying.... -
- This is the patch to fix this problem: https://github.com/ruby/ruby/pull/4769
08/24/2021
-
- What is observed in ppc64le CI (pathnames edited for readability):
```
-- C level backtrace information -------------------------------------------
/home/xxx/ruby/ruby(rb_vm_bugreport+0x198) [0x6430d199028] vm_dump.c:759
/home/xxx/... - 03:36 PM Revision 71662466 (git): * 2021-08-25 [ci skip]
-
03:34 PM Revision 14a9c364 (git): Fix warnings in test_optimization.rb
- These were introduced in the test for tracing optimized methods
added in 48c8df9e0eb295af06d593ce37ce1933c0ee1d90. -
- Applied in changeset commit:git|bbedd29b6e98ef6e3fc2ce2b358d2b509b7cd1bb.
----------
[Bug #18117] Fix Ractor race condition with GC
rb_objspace_reachable_objects_from requires that the GC not be active.
Since the Ractor barrier is not ... -
- rb_objspace_reachable_objects_from requires that the GC not be active.
Since the Ractor barrier is not executed for incremental sweeping,
Ractor may call rb_objspace_reachable_objects_from after sweeping
has started to share objects. Thi... -
- What I noticed in the description is that "C level backtrace information" is not shown, which suggests that the functions in `addr2line.c` are never called from `rb_vm_bugreport` (or its offspring). Thus I suppose #17052 is irrelevant to...
-
11:45 AM Bug #16492: TestBugReporter#test_bug_reporter_add test failures
- Perhaps this issue is relevant with #17052 .
-
- The reason of the unexpected error is, the block parameter of `define_method` is un-shareable.
I created a patch to explain it: https://github.com/ruby/ruby/pull/4771 -
- I noticed gcc-11 LTO and clang-14 LTO cases are added to the CI. Thanks @nobu!
https://github.com/ruby/ruby/commit/29dd70492eb3e3f0018c01fb3fa4ce77029a208f
https://github.com/ruby/ruby/commit/244fc65d1c58364b8b9817e07983ae651272b033
-
- As a follow-up of this report, I found that the same thing happens when:
* a Ractor accepts connections, then send connections to other Ractor workers
* Ractor workers read/write from/to connections, then close those
This is a very ... -
- https://github.com/ruby/securerandom/commit/de47532707
-
- This reverts commit 22deda43cb98aa3cee48d0bebbff7c4db1d7652a.
It was incomplete. Sorry! -
09:02 AM Revision 80597f25 (git): [ruby/un] gemspec: Explicitly list 0 executables
- This gem exposes no executables, so this makes that clear to the reader.
https://github.com/ruby/un/commit/012c298e12 -
- https://github.com/ruby/un/commit/762c38c176
-
- https://github.com/ruby/un/commit/8e0d6f62cb
-
- https://github.com/ruby/un/commit/4d080eeec0
-
- 08:55 AM Revision 2be84afe (git): .github/workflows/compilers.yml: Specify a container running user as root.
- Explicitly specify the root as a user to run tests in the containers.
Coming new ruby/ruby-ci-image images are required to run the container as
a regular user by default, while the root user is required to run the
compilers.yml. Add `id... -
- Formerly, TypeProf was tested with the latest RBS code during
`make test-bundled-gems`. However, when a new version of rbs is
released, and if it is incompatible with TypeProf,
`make test-bundled-gems` starts failing, which was annoying.... -
- `ErrorHighlight.formatter` fetches a default (or configured) exception formatter from the module variable, and if it occurs in Ractor, it causes additional exceptions.
```
/Users/tagomoris/.rbenv/versions/3.1.0-dev/lib/ruby/3.1.0/err... -
- * Simplify libc_so and libm_so
If nil, no need to set to nil.
* Get rid of repeating inversions
https://github.com/ruby/fiddle/commit/4323e689d8 -
07:18 AM Revision 0f1e8f38 (git): [ruby/fiddle] Improve "offsetof" calculations (https://github.com/ruby/fiddle/pull/90)
- I need to get the offset of members inside sub structures. This patch
adds sub-structure offset support for structs.
https://github.com/ruby/fiddle/commit/cf78eddbb6 - 01:52 AM Revision 8d2af51a (git): netbsd coroutine uses assembly instead and little build fix.
-
- Applied in changeset commit:git|c38c2d8ee26c70300e2e0a4eec867273a8520579.
----------
Moved exported symbols in internal/util.h to ruby/util.h
[Feature #18051] -
-
-
- [Feature #18051]
-
12:23 AM Revision 4794a8a7 (git): Add stat_time function
08/23/2021
-
10:08 PM Bug #18129 (Closed): Ractor-incompatible global variables can be accessed through alias
- Some global variables are not allowed in non-main Ractors. For example, the following code does not run:
``` ruby
Ractor.new do
$/
end.take
#=> can not access global variables $/ from non-main Ractors (Ractor::IsolationError)
```... -
- In order to avoid data races, Ractor is not supposed to allow class variables. For example, the following code does not run:
``` ruby
class C
def initialize
@@var = 1
end
end
Ractor.new do
C.new
end.take
#=> can not a... -
- **Background**
When the Singleton module (from the Singleton library) is included in a class, that class will have only one instance. Since the instance can only be in one Ractor at once, Singleton is not Ractor-compatible. For example,... - 06:58 PM Revision 70abda1a (git): * 2021-08-24 [ci skip]
-
- Pointers may be large to the point where intptr_t would be negative.
This is problematic when doing comparisons of pointers. -
xtkoba (Tee KOBAYASHI) wrote in #note-31:
> I might have been misleading. The patch in #note-9 is necessary for correctly reading `DW_FORM_ref_addr` values in `.debug_info` sections, even if those values are kept uninterpreted by the ...-
- I might have been misleading. The patch in #note-9 is necessary for correctly reading `DW_FORM_ref_addr` values in `.debug_info` sections, even if those values are kept uninterpreted by the patch in #note-11. Both patches have been merge...
-
- xtkoba (Tee KOBAYASHI) wrote in #note-29:
> The behavior seems as if the patch in #note-9
Should it be applied? Based on the #note-10, #note-11 and #note-12, I thought that one is obsolete. That might be misunderstanding on my side, ... -
- The behavior seems as if the patch in #note-9 (https://bugs.ruby-lang.org/attachments/8974) were not applied. If I reverse-apply this patch, yes I get the following output:
```
1300: debug_info 0x54 Abbrev Number 101 not found
Abort... -
- Just FTR, these are the build options:
~~~
gcc -O2 -flto=auto -ffat-lto-objects -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -specs=/usr/lib/rpm/redhat/re... -
- Byte 0x54 of the `.debug_info` is the 3rd byte of
```
<52> DW_AT_abstract_origin: <0x6a6c3a>
```
of the form `DW_FORM_ref_addr` which is in little endian 0x6a == 106 (and also the same value when interpreted as ULEB128). So... -
- ~~~
(gdb) c
Continuing.
-e:1: [BUG] Segmentation fault at 0x590fb15c00000020
ruby 3.0.2p107 (2021-07-07 revision 0db68f0233) [powerpc64le-linux]
-- Control frame information -----------------------------------------------
c:0003 ... -
- This reverts commits 48ff7a9f3e47bffb3e4d067a12ba9b936261caa0
and b2e2cf2dedd104acad8610721db5e4d341f135ef because it is causing
crashes in SPARC solaris and i386 debian. -
- Applied in changeset commit:git|48ff7a9f3e47bffb3e4d067a12ba9b936261caa0.
----------
[Feature #18045] Remove T_PAYLOAD
This commit removes T_PAYLOAD since the new VWA implementation no longer
requires T_PAYLOAD types.
Co-authored-by: ... -
- This commits implements size classes in the GC for the Variable Width
Allocation feature. Unless `USE_RVARGC` compile flag is set, only a
single size class is created, maintaining current behaviour. See the
redmine ticket for more detail... -
- This commit removes T_PAYLOAD since the new VWA implementation no longer
requires T_PAYLOAD types.
Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org> -
- The following code occasionally dumps core.
```
3000.times{Thread.new{}}
sleep 2.99
```
```
$ ruby test.rb
[BUG] Segmentation fault at 0x0000000000000440
ruby 3.0.1p64 (2021-04-05 revision 0fb782ee38) [x86_64-linux]
$
```... -
-
01:59 AM Feature #17847 (Closed): `ruby -run -e httpd` displays URL
- > Alternatively this could be part of the default output when starting WEBrick,
That's a good idea. However, since the definition of the port number output is in the HTTPServer superclass WEBrick::GenericServer, the amount of change i... - 12:59 AM Revision 116e52f3 (git): * 2021-08-23 [ci skip]
- 12:59 AM Revision 689962ea (git): improves message. thanks to eregon
- 12:59 AM Revision f18e1752 (git): Display httpd URLs supporting HTTPS
- 12:59 AM Revision 8361675e (git): Display httpd URLs
08/22/2021
-
-
- https://github.com/ruby/date/commit/7e1ffbf568
-
-
- Call `IOSpecs.io_fixture` with the default encoding explicitly.
`IOSpecs.closed_io` calls the method without optional `mode` which
is set to UTF-8 by default, while the default external encoding
depends on the locale environment variables. -
-
01:33 AM Bug #18077 (Closed): Marshal.dump(closed_io) raises IOError instead of TypeError - Applied in changeset commit:git|6594623f623a0982da62cc105094da0701d499da.
----------
Fix Marshal.dump(closed_io) to raise TypeError and allow encoding on closed IO
Mashalling a closed IO object raised "closed stream (IOError)" before i... - 01:33 AM Revision 6594623f (git): Fix Marshal.dump(closed_io) to raise TypeError and allow encoding on closed IO
- Mashalling a closed IO object raised "closed stream (IOError)" before instead of TypeError.
This changes IO#(in|ex)ternal_encoding to still return the encoding even if the underlying FD is closed.
Fixes bug #18077
08/21/2021
-
09:19 PM Bug #18077: Marshal.dump(closed_io) raises IOError instead of TypeError
- I should have read the PR first.
Yes, removing the old example and adding the new one with a version guard is perfect here :)
That allows other implementations & older versions to potentially use the new behavior already, and there is ... -
- larskanis (Lars Kanis) wrote in #note-5:
> @eregon This was my first though as well, but the current behavior is already defined in ruby-spec.
We can change it in ruby/spec (adding version guards and adding the new expectation).
And... -
- I removed the old example from https://github.com/ruby/ruby/pull/4758 .
-
- Since this is a bug report, I think that the old example in rubyspec should be removed.
-
05:47 PM Revision c527d278 (git): Bundle RBS 1.5.1 (#4760)
-
- Applied in changeset commit:git|48c8df9e0eb295af06d593ce37ce1933c0ee1d90.
----------
Allow tracing of optimized methods
This updates the trace instructions to directly dispatch to
opt_send_without_block. So this should cause no slowdo... - 05:15 PM Revision ad99edc7 (git): * 2021-08-22 [ci skip]
-
- This updates the trace instructions to directly dispatch to
opt_send_without_block. So this should cause no slowdown in
non-trace mode.
To enable the tracing of the optimized methods, RUBY_EVENT_C_CALL
and RUBY_EVENT_C_RETURN are added... -
09:31 AM Misc #18125 (Closed): A strange behavior when same name variable/method coexist issue.
- Local variables are defined when they appear in assignments, thus
```ruby
deploy_to = "#{deploy_to} new place"
# ^ assignment ^local variable
```
To avoid ambiguity, use parenthesis.
```ruby
deploy_to = "#{deploy_to()}... -
08:28 AM Misc #18125: A strange behavior when same name variable/method coexist issue.
- It is excepted because of variable hoisting.
Same reason why `foo = 1 unless foo` works.
-
08:07 AM Misc #18125 (Closed): A strange behavior when same name variable/method coexist issue.
- Following is a example.
## Reproduce
```rb
def deploy_to
"deploy_to"
end
deploy_to = "#{deploy_to} new place"
p defined? deploy_to # => local_varible
p deploy_to # => " new place"
```
## What we expected.
```rb
... -
-
-
- > It can affect performance on some cases.
But less than `GC::Profiler`, so for users of this feature it would be a performance win.
> ...
I don't think it's a big problem if it isn't perfectly accurate.
> Adding flag like GC.m... -
06:27 AM Bug #17643: Ruby 3 embedded - no GC methods?
- I don't want to do either of those things. Instead I would like to embed Ruby into a C++ program and then use it run tests.
Specifically, this is for Rice. The embed Ruby code:
https://github.com/jasonroelofs/rice/blob/master/test/... -
- No arguments means reading the script from stdin.
You need to give a script name or an `-e` option. - 02:22 AM Revision e2b2b577 (git): * 2021-08-21 [ci skip]
-
- `rb_cData` has been deprecated for years.
-
12:52 AM Feature #18083: Capture error in ensure block.
- > A few places, but in relative usages probably 1/1000 of begin/rescue or ensure/end patterns, right?
I scanned top 10,000 gems by download count:
Checking for "$!":
Found 459 gems with this pattern.
Found 2624 instances of the pattern...