Activity
From 10/03/2021 to 10/09/2021
10/09/2021
- 04:24 PM Revision 6678bb66 (git): * 2021-10-10 [ci skip]
-
04:20 PM Revision 7f3786c3 (git): Use an isolated class
- The number of all instances of `Class` may be affected by GC-able
anonymous classes created by other tests. -
02:29 PM Revision 5805b147 (git): [ruby/irb] Version 1.3.8.pre.11
- https://github.com/ruby/irb/commit/49b9f644c2
-
- https://github.com/ruby/irb/commit/4f686afbca
-
- https://github.com/ruby/reline/commit/de858ee1ac
-
- https://github.com/ruby/reline/commit/6a0d0ada94
-
- https://github.com/ruby/reline/commit/cff83e68f8
-
- The setting to use '::' instead of block elements.
https://github.com/ruby/reline/commit/4afbc98e2c -
- https://github.com/ruby/reline/commit/93bc9b5277
-
- * Separate exception classes to be rescued or reraised
* Use the filtered backtrace in the failure message
* Raise a new `AssertionFailedError` with the original backtrace -
- I agree with @jeremyevans0 on that this is not a bug.
Feature requests need the approval by Matz to merge. -
02:40 AM Feature #18242: Parser makes multiple assignment sad in confusing way
- Attached is an improved driver. Again, every test case works with nobu's branch.
-
- @nobu I built your PR branch and confirmed that this new parsing works great with the `driver.rb` attached.
Will this patch have to wait for the 3.1.0 release? Sorry I'm not sure how the whole patching and backporting processes work. ... -
- * [Feature #18242] Allow multiple assignment in logical expression
* `1 < 2 and a, b = 2, 1` and so on. -
-
- Merge `Test::Unit::CoreAssertions#message`.
-
06:38 AM Bug #17507: Regexp capture groups ignored sometimes in some multithreaded environments (possible race condition)
- ruby_3_0 cfad0583eb18bb4505f28ee5cfab703a6b9987be merged revision(s) abc0304cb28cb9dcc3476993bc487884c139fd11.
-
- ruby_3_0 2c947e74a0a11fe6c54253c15224dc80054c62a2 merged revision(s) 60d0421ca861944459f52292d65dbf0ece26e38a,b6534691a16d751d59fc572d5dddebcaeb21f007,409dbc951b9875d27bd73748c88e15386473cffb,842b0008c132dd587f09766a228041afb7fed24f.
-
- ruby_3_0 fe9d33beb78d5c7932a5c2ca3953045c0ae751d5 merged revision(s) 89242279e61b023a81c58065c62a82de8829d0b3,529fc204af84f825f98f83c34b004acbaa802615.
-
- Avoid race condition in Regexp#match
In certain conditions, Regexp#match could return a MatchData with
missing captures. This seems to require at the least, multiple
threads calling a method that calls the same ... -
- Fix the encoding of loaded feature names [Bug #18191]
The feature names loaded from the default load paths should also
be in the file system encoding.
---
ruby.c | 12 +++++++++++-
... -
- Marshal.load: do not call the proc until strings have their encoding
Ref: https://bugs.ruby-lang.org/issues/18141
---
marshal.c | 7 +++-
spec/ruby/core/marshal/shared/load.r... -
- Similar to `Test::Unit::CoreAssertions::AllFailures#foreach`.
-
12:07 AM Revision 7e506716 (git): Newly generated gems require Ruby 2.6.0
- In 2021, Ruby 2.5 and older are EOL.
We can set the default required Ruby version to 2.6.0 to
encourage people to use newer Ruby.
If the command is executed with old Ruby, it falls back to 2.3.0.
It's still possible to create a gem for o...
10/08/2021
-
11:05 PM Revision 91f794b5 (git): [rubygems/rubygems] I see no harm in this, allow it
- https://github.com/rubygems/rubygems/commit/01feb40283
-
- https://github.com/rubygems/rubygems/commit/12af03d32f
-
- Unless the `:branch` option is passed.
It's more efficient, and it results in less hardcoding of "master".
https://github.com/rubygems/rubygems/commit/aa5c3409ab -
- For efficiency.
https://github.com/rubygems/rubygems/commit/979d1634dd -
- This doesn't affect the outcome of the test, but it makes the `git
commit` command inside `update_git` not fail because of not having
anything to commit.
https://github.com/rubygems/rubygems/commit/ad0160ed97 -
- We want to update the original repo, not a fresh one.
This went undetected because the `git commit` command inside the
`update_git` method ignores failures, and in this case it was failing
because all files are untracked in the new repo... -
11:04 PM Revision d03d122b (git): [rubygems/rubygems] Improve performance of Specification#missing_extensions?
- https://github.com/rubygems/rubygems/commit/90c1919f94
- 11:03 PM Revision ad92651d (git): [rubygems/rubygems] Add glob infomation to Bundler::Source::Git#to_s
- The glob information was not specified in the string representation for
a source, which led to non-deterministic behaviour when generating the
lockfile, since sources are sorted by this value.
https://github.com/rubygems/rubygems/commit... -
11:02 PM Bug #18000 (Closed): have_library doesn't work when ruby is compiled with --disable-shared --disable-install-static-library
- Applied in changeset commit:git|06c3e80611b81ec8f251957328486e9b6dd18d3b.
----------
Do not allow configuration where neither static or shared library is installed
Fixes [Bug #18000] -
- Fixes [Bug #18000]
-
10:07 PM Feature #3848 (Closed): Using http basic authentication for FTP with Open URI
-
-
- Applied in changeset commit:git|08759edea8fb75d46c3e75217e6613465426a0d2.
----------
Remove autoload for constant if the autoload fails
Previously, if an autoload failed (the file was loaded, but the
constant was not defined by the aut... - 09:54 PM Revision b4afedb0 (git): * 2021-10-09 [ci skip]
-
- Previously, if an autoload failed (the file was loaded, but the
constant was not defined by the autoloaded file). Ruby will try
to autoload again if you delete the autoloaded file from
$LOADED_FEATURES. With this change, the autoload an... -
-
-
- nobu (Nobuyoshi Nakada) wrote in #note-3:
> That patch seems not including `1 < 2 and a = 1, 2`.
> ...
Thank you @nobu ! -
03:38 PM Bug #18247 (Closed): weird results for `Array#slice` or `Array#[]` with argument of type `Enumerator::ArithmeticSequence`
- There are weird results for `Array#slice` or `Array#[]` with argument of type `Enumerator::ArithmeticSequence`.
Particularly most results with negative steps are weird. For example:
```ruby
@array = [0, 1, 2, 3, 4, 5]
@array.sl... -
10:48 AM Bug #18244: Unexpected errors output order
- Glad to have helped :)
And I am sorry if my reply sounded harsh.
I think we can close this, I'll let @nobu decide.
Although I'm not against nobu's patch, I think it is reasonably expected behavior when using `$stderr.sync = false`. -
05:57 AM Bug #18244: Unexpected errors output order
- >I believe the real issue here is to $stderr.sync = false, why do you do that?
> ...
Having buffered `$stdout` I thought that `$stderr.sync = false` will make `$stderr` to be in sync with `$stdout`. Now I understand that it was not the ... -
- https://github.com/ruby/reline/commit/69e365edcc
-
-
06:40 AM Bug #10416: Create mechanism for updating of Unicode data files downstreams when we want
- jeremyevans0 (Jeremy Evans) wrote in #note-7:
> @duerst Do you know if this is still in issue in the master branch?
- I suspect it is still "an issue", i.e. it still happens.
- Nobody has complained about it, and so it may be that i... -
- @duerst Do you know if this is still in issue in the master branch?
-
- Also, check if a suffix is empty, to guarantee the assumption of
`onigenc_get_left_adjust_char_head` that `*s` is always accessible,
even in the case of `SHARABLE_MIDDLE_SUBSTRING`. -
-
- Previous fix didn't show the cursor or dialogs. This allows the
dialogs to continue to work.
https://github.com/ruby/reline/commit/651c5f63cc -
- Fixes a crash in IRB if a dialog is displayed and the default
external encoding is not UTF-8:
/home/jeremy/tmp/reline/lib/reline/line_editor.rb:731:in `write': U+2588 from UTF-8 to US-ASCII (Encoding::UndefinedConversionError)
https://... -
- But pending them now because they need dummy document data to show doc.
https://github.com/ruby/irb/commit/ac471ee14e -
- https://github.com/ruby/irb/commit/5df6e1f027
-
- https://github.com/ruby/irb/commit/f34da7fa04
-
- https://github.com/ruby/irb/commit/a23a88b8c9
-
- https://github.com/ruby/irb/commit/4d7cefcaa4
-
- https://github.com/ruby/reline/commit/44cd35e65d
-
- https://github.com/ruby/reline/commit/972cc993ca
-
- https://github.com/ruby/reline/commit/996bfec64b
-
12:59 AM Bug #18245 (Third Party's Issue): CSV Parser, buffer overflow issue with very specific data
- Could you open an issue on https://github.com/ruby/csv ? We want to track csv gem problem on https://github.com/ruby/csv .
-
10/07/2021
-
- I believe the real issue here is to `$stderr.sync = false`, why do you do that?
That's like explicitly breaking what you are asking.
If it's for performance because you are writing a lot to stderr, then why are you writing so much to... -
- > Why mixing `stdout` and `stderr`?
I have a script that is running periodically in the background on another machine. Sometimes I visit that machine and review logs.
My goal is to see a full story in the log, the same as we can see... -
- It sounds hard anyway regardless `stderr` buffering.
Why mixing `stdout` and `stderr`?
https://github.com/ruby/ruby/pull/4944 -
- > Do you have any reason to make `stderr` unbuffered?
I do add records to stdout within a cycle, so it is better to have it buffered for better overall performance. Also, I have few parallel workers in threads with such cycles. As a r... -
- Of course it's possible to make flushing `stderr` at error handling, but it also can make a chance of another exception.
Do you have any reason to make `stderr` unbuffered? -
- Having this code:
```ruby
# $stderr.sync = false
warn 'warn'
$stderr.puts 'err'
fail
```
When I run it like this `ruby test.rb 2>err.log`, output is ok:
```
warn
err
test.rb:3:in `<main>': unhandled exception
```
But when ... -
-
10:40 PM Bug #18232 (Closed): Ractor.make_shareable is broken in code loaded with RubyVM::InstructionSequence.load_from_binary
- Applied in changeset commit:git|217df51f0e5d9824ed712a4d175f555d932e44d8.
----------
Dump outer variables tables when dumping an iseq to binary
This commit dumps the outer variables table when dumping an iseq to
binary. This fixes a c... -
- This commit dumps the outer variables table when dumping an iseq to
binary. This fixes a case where Ractors aren't able to tell what outer
variables belong to a lambda after the lambda is loaded via ISeq.load_from_binary
[Bug #18232] [... -
-
-
-
-
-
- Support for this was added in Ruby 3.0.
-
- `RBOOL` was recently introduced for this purpose.
-
-
- SortedSet was removed in commit:a3db08d7b6ff119223f77e3df00b4f6deac971e2
-
08:04 PM Feature #17849: Fix Timeout.timeout so that it can be used in threaded Web servers
- The fiber scheduler redefines `Timeout` to only well defined wait points.
We could also do the same for other options likes `Thread#raise` etc. -
12:56 PM Feature #17849: Fix Timeout.timeout so that it can be used in threaded Web servers
- mame (Yusuke Endoh) wrote in #note-14:
> Just FYI. According to @ko1, masking all exceptions in an ensure clause is possible in pure Ruby, as long as you use MRI.
> ...
This is incorrect code. The `handle_interrupt` code must go outsid... -
- The method you are looking for is `!` and not `!@`. The reason `false.send(:!@)` works is that `:!@` is parsed as `:!`, that's why the unquoted version works but the quoted version does not. This was determined to be a feature and not ...
-
05:36 PM Bug #18246 (Closed): send does not work for unary ! operator when operator isn't a literal symbol
- ```ruby
1.send(:+, 1) #=> 2
1.send(:"+", 1) #=> 2
1.send(:-@) #=> -1
1.send(:"-@") #=> -1
false.send(:!@) #=> true
false.send(:"!@") #>
# undefined method `!@' for false:FalseClass (NoMethodError)
# Did you mean? !=
... -
-
- This bug happens when a multibyte row separator (such as `\r\n`) is split when reading a chunk. I've submitted a pull request to fix the handling of `\r\n` row separators: https://github.com/ruby/csv/pull/221. It's a suboptimal fix as ...
-
08:23 AM Bug #18245 (Third Party's Issue): CSV Parser, buffer overflow issue with very specific data
- This may not fall into guidelines since it's a very specific issue, but I have exhausted every avenue of this to be a File issue.
Ruby (2.6.6) native CSV parser crashes on a specific file. I have tried reproducing the exact set of byt... -
- That patch seems not including `1 < 2 and a = 1, 2`.
https://github.com/ruby/ruby/pull/4945 - 03:15 PM Revision cb5a3b19 (git): * 2021-10-08 [ci skip]
-
-
02:55 PM Revision 80e2242d (git): [ruby/irb] Update descriptions of methods
- From Reidline to Reline
Update description used in take_corresponding_syntax_to_kw_do and is_the_in_correspond_to_a_for methods
Use possessive noun correctly
Second element
https://github.com/ruby/irb/commit/4fa9714d6f -
- Applied in changeset commit:git|896626fa02f00831de5db2e3e75aa349a5a84102.
----------
[ruby/irb] Make IRB::Context#exit call super if UncaughtThrowError
Fixes calling exit after binding.irb.
Fixes [Bug #18234]
https://github.com/ruby/... -
- Fixes calling exit after binding.irb.
Fixes [Bug #18234]
https://github.com/ruby/irb/commit/4ea8d376f2 -
-
-
09:34 AM Revision ae5dffd6 (git): Followed up bd6e1a0f0883dba7b02f30cefe5ebec96d02cb90
-
- These are supported by Ruby's socket library if the operating system
supports zone indentifiers, so they should be supported by ipaddr.
See RFCs 4007 and 6874 for additional information.
Implements Ruby Feature #10911
https://github.co... - 09:22 AM Revision 74ed881e (git): [ruby/ipaddr] Add netmask method that returns net mask as string.
- https://github.com/ruby/ipaddr/commit/283d16f3a3
-
- https://github.com/ruby/ipaddr/commit/f49d2d49a4
-
- https://bugs.ruby-lang.org/issues/15832
https://github.com/ruby/ipaddr/commit/de9805d6fa -
- Instead of raising a new exception with a modified message, just
use the correct message to begin with. This avoids the issue with
both exceptions being displayed at error exit.
https://github.com/ruby/ipaddr/commit/09edfd4a7f -
08:46 AM Revision b23fba91 (git): [ruby/ipaddr] Removing superfluos assingments & return
- Also adding test for ntop
https://github.com/ruby/ipaddr/commit/0ba16cca10 -
06:18 AM Revision 323e7a67 (git): Merge if statement
-
03:11 AM Feature #12075: some container#nonempty?
- > Intuitively, "x.size?" is more likely interpreted as "is the size defined for x?" rather than "is the size of x greater than 0?"
When one knew `size?` always comes with `size`, one should see that asking that question does not make ... -
12:51 AM Feature #12075: some container#nonempty?
- knu (Akinori MUSHA) wrote in #note-33:
> [...]`size?` [...] explicitly says it would check the size, which means it is essentially equivalent to `size > 0`
While this might make sense in JavaScript, given that `0` is truthy in Ruby, I d... -
01:40 AM Revision 273673ae (git): test/ruby/test_process (test_no_curdir): tentatively skipped on Solaris
- The test gets stuck on Solaris CI.
http://rubyci.s3.amazonaws.com/solaris11-gcc/ruby-master/log/20211006T050007Z.fail.html.gz#test-all
```
[14558/21042] TestProcess#test_no_curdirtimeout: output interval exceeds 1800.0 seconds.
timeout:... -
12:06 AM Revision 7caeead3 (git): Accommondate earlier reviews of RDoc for Enumerable (#4943)
-
12:00 AM Feature #14794: Primitive arrays (Ruby 3x3)
- saw from https://chrisseaton.com/truffleruby/rubykaigi21/ - the shape of ruby objects. Maybe if we can get this implemented it would help out with this. Or use collection storage strategies to implement this. https://tratt.net/laurie/re...
10/06/2021
-
- I've attached a patch that fixes this (for me) along with a test. I had to change the information that we dump when dumping an ISeq, so maybe we need to change the binary format version number or something? I'm not sure. Anyway there ...
-
- It looks like the [outer variables list](https://github.com/ruby/ruby/blob/e4d85d3a2d0c7d4a1abf981fb1ba557e21cad968/vm.c#L1124) is being lost when we dump the iseq to binary. AFAICT, that list is only created at compile time. I'll try ...
-
- (this comment was moved to https://bugs.ruby-lang.org/issues/18243, replied to the wrong issue)
-
- FYI the way it currently works for OpenStruct seems buggy on the Ractor side: #18243
-
09:22 AM Bug #18232: Ractor.make_shareable is broken in code loaded with RubyVM::InstructionSequence.load_from_binary
- Because [OpenStruct was recently changed to use `make_shareable`](https://github.com/ruby/ostruct/pull/29), this bug suddenly became much more problematic. We actually hit it in production.
-
-
- jeremyevans0 (Jeremy Evans) wrote in #note-1:
> [...] Attached is a patch that implements the necessary parser rules, without causing any conflicts. However, the patch doesn't include the parts needed to make ripper work, as I couldn't... -
- The parser is currently not setup to handle this, so I don't think it is a bug. You would have to add new parser rules to allow it, so I would consider this a feature request and not a bug. Attached is a patch that implements the neces...
-
- Example:
``` ruby
a, b = 2, 1 if 1 < 2 # Works
a, b = [2, 1] if 1 < 2 # Works
(a, b) = 2, 1 if 1 < 2 # Works
(a, b) = [2, 1] if 1 < 2 # Works
(a, b = [2, 1]) if 1 < 2 # Works
a, b = 2, 1 unless 2 < ... -
- Treated:
#sum
#uniq
#compact -
-
-
-
-
- This was fixed in Ruby 2.2:
```
$ ruby22 -ve 's = "foo"; class << s; def =~ (rhs); raise "a"; end; end; s =~ /foo/'
ruby 2.2.10p489 (2018-03-28 revision 63023) [x86_64-openbsd]
-e:1:in `=~': a (RuntimeError)
from -e:1:in `... -
-
- This feature was implemented as `transform_keys`, `transform_values`, and `to_h`.
-
- This feature was implemented as `transform_keys`, `transform_values`, and `to_h`.
-
- This was implemented by `Hash#to_h` taking a block starting in Ruby 2.6.
-
07:38 PM Bug #18023 (Closed): Ractor#make_shareable changes the values of closured shareable objects
- Applied in changeset commit:git|76228191474c76810043b294a74bbb2f1808b3d9.
----------
Fix Ractor.make_shareable changing locals for Procs
env_copy() uses rb_ary_delete_at() with a loop counting up while
iterating through the list of rea... - 07:38 PM Revision 121424a0 (git): * 2021-10-07 [ci skip]
-
- env_copy() uses rb_ary_delete_at() with a loop counting up while
iterating through the list of read only locals. rb_ary_delete_at() can
shift elements in the array to an index lesser than the loop index,
causing locals to be missed and s... -
- Smaller repro:
```
$ ruby -e 'Object.new.instance_eval { p object_id; proc = Ractor.make_shareable -> { self }; Ractor.new(proc) { |c| p c.call.object_id }.take }'
60
<internal:ractor>:267: warning: Ractor is experimental, and the be... -
- ```ruby
class C
attr_accessor :foo
def setter_proc
Ractor.make_shareable(-> v { @foo = v })
end
end
c = C.new
c.foo = 1
p c
proc = c.setter_proc
p c.frozen?
Ractor.new(proc) { |s| s.call(42) }.take
p c
```
give... -
- This proposal has basically been implemented, with single-line pattern matching and valueless hash creation syntax:
```ruby
john = {name: "John", last: "Smith"}
john => {name:, last:}
name # => "John"
last # => "Smith"
p({name:, ... -
- I submitted a pull request to irb to fix this by calling super if UncaughtThrowError is raised by IRB::Context#exit: https://github.com/ruby/irb/pull/298
-
- Probably not, considering that some classes I named earlier, File::Stat and Enumerator, only have #size and no #length.
Adding #length? in itself would be fine, but that'd make you feel awkward not to add #length also. 😅 -
11:19 AM Feature #12075: some container#nonempty?
- If `size?` will be chosen, should we also add `length?` as `size` and `length` are aliases?
-
-
- `rb_encoding` is defined as `const OnigEncodingType`.
Fix lots of C4114 warnings for each files by MSVC. -
-
- Ruby cannot guarantee the resolutions of underlying filesystems.
10/05/2021
-
- Treats:
#slice_after
#slice_when
#chunk_while -
-
- Fixes [Feature #18148]
When set, all the loaded objects are returned as frozen.
If a proc is provided, it is called with the objects already frozen. -
- Applied in changeset commit:git|afcbb501ac17ba2ad5370ada5fd26e8dda9a5aaa.
----------
marshal.c Marshal.load accepts a freeze: true option.
Fixes [Feature #18148]
When set, all the loaded objects are returned as frozen.
If a proc is p... -
- * Enhanced RDoc for Enumerable#slice_before
* Enhanced RDoc for Enumerable#slice_before -
- Applied in changeset commit:git|c6706f15af123bdbb3b39a21903d85c78462d047.
----------
Fix documentation for String#{<<,concat,prepend}
These methods mutate and return the receiver, they don't create
and return a new string.
Fixes [Bug ... -
- This is a documentation bug. Both String#<< and String#concat mutate and return the receiver. I'll fix this shortly.
-
03:22 PM Bug #18241: Question on String Concatentation Documentation
- edit: to be clear, I'm making a distinction between "new string" in the contents-sense, and "new string" in the memory location sense.
For example, the `<<` operation on `Array` inserts an array entry and returns `self`, which is why ... -
- Hello,
I noticed that the documentation for `<<` and `concat` both state that the result of each operation is a new string.
However, while doing some experimenting myself I noticed that this doesn't seem to be the case. The result ... - 03:41 PM Revision f883f550 (git): * 2021-10-06 [ci skip]
- 03:40 PM Revision 1bf3f3f4 (git): * remove trailing spaces. [ci skip]
-
- These methods mutate and return the receiver, they don't create
and return a new string.
Fixes [Bug #18241] -
-
03:33 PM Bug #18240: Random segfault on a local Rails server
Thanks Jeremy for your answer. We're gonna try to disable the C extensions one by one to identify if one behaves badly.
I think the ticket can be marked as closed, we will publish our result here.-
- Unfortunately, I don't think there is enough information provided to help diagnose the specific cause of the issue (the general cause is likely to be memory corruption). The following 3rd party C-extensions are in use:
```
nio4r
bi... -
- Hello
I am having random segfault locally when rails a basic rails app.
The control frame is weird:
| -- Control frame information -----------------------------------------------
web | c:0073 p:---- s:0470 e:000469 CFUNC... -
- Regarding ABI compatibility, AFAIK CRuby doesn't care about it for development versions.
I.e., conceptually we can think of every commit potentially changing the ABI but the ABI version number (`RbConfig::CONFIG['ruby_version']`) is nev... -
01:08 PM Feature #18239: Variable Width Allocation: Strings
- Hi @duerst. Thank you for adding it to the developer meeting.
I understand your concern for gem incompatibility. Since the public-facing header files often include internal implementation details, we don't expect this to be the last t... -
02:22 PM Revision f2ca6526 (git): debug.gem v1.2.4
- This version fixes test failures on MacOS enviornment.
-
- https://github.com/ruby/reline/commit/22359d50ab
-
- https://github.com/ruby/reline/commit/5f4a75c7a0
-
- So testing is no longer necessary.
https://github.com/ruby/reline/commit/c59589548b -
- https://github.com/ruby/reline/commit/df2a1b4e08
-
-
- I dislike unnatural casts.
-
- The former uses the current location, while the latter takes a
non-null location. -
-
-
- When Bison reports "memory exhausted", it means the parser stack
depth reached the limit `YYMAXDEPTH` which is defaulted to 10_000,
but not memory allocation failed. -
- This version uses tempdir instead of homedir to store UNIX domain
socket. -
05:18 AM Revision f032c09b (git): rb_enc_left_char_head(): take void*
- Nobu doesn't like (char*) cast.
-
- These functions assume ASCII compatibility. That has to be ensured in
their caller. -
-
- Less macros == huge win.
-
- 2,291 lines are too much! include/ruby/encoding.h became the biggest
header file once it had doxygen comments. Let us split it into smaller
parts, so that we can better organise their contents. -
- My bad. The document is clearly broken. Maybe I pressed my delete key
too much. [ci skip] -
- Why did they even exist?
-
- After the change to use realpath on loaded features, Solaris CI
started failing in test_no_curdir (which tests behavior for running
ruby without a working directory).
I was able to trace the problem to the following call chain:
rb_call... - 04:06 AM Revision 16ce578d (git): Enable inline assembly of SET_MACHINE_STACK_END on AIX
10/04/2021
-
- This is impressive work. I haven't time to study all the details, in particular the benchmark results.
But I'm somewhat worried by:
> Gems with native extensions built with VWA are not compatible with non-VWA built gems (and vice v... -
- Ah, fool question. Sorry. It was arguments on rdoc benchmark (I misunderstood that it is for general characteristics).
-
- Yes, [rdoc generation](#rdoc-generation). Speed is 0.95x but memory usage dropped to 0.81x (0.89x for jemalloc).
-
- > The performance decrease is small but the memory decrease is large.
Sorry, which results do you say? rdoc generation?
-
- Your concerns about memory usage and performance are all valid. A solution is we could make VWA enabled by default but only default to 1 size pool (instead of the 4 size pools right now) and make the number of size pools configurable thr...
-
- > I can't make good reason why VWA is slower than FWA on your benchmark.
The only benchmark above where VWA is slower than FWA is rdoc generation. The performance decrease is small but the memory decrease is large. This is because VWA... -
- * Performance:
* P4. `GC.compact` can be slower than FWA because it should be copy the body. But maybe it is not a matter. -
- Thank you for your reply.
peterzhu2118 (Peter Zhu) wrote in #note-3:
> - The RString layout does not change much. The only big change is that the length for embedded strings is moved from using 5 bits in the header to a field in the ... -
- Sorry for incremental questions.
Why do we need to enable it now? Current benchmark you showed illustrate not so difference.
"On a live, production Shopify service, we see no significant differences in response times. However, VWA has l... -
- Hi @ko1, thank you for the feedback.
- The RString layout does not change much. The only big change is that the length for embedded strings is moved from using 5 bits in the header to a field in the RString. This is because the 5 bits... -
- ko1 (Koichi Sasada) wrote in #note-1:
> * Could you explain the variable size page strategy? https://bugs.ruby-lang.org/issues/18045 shows 4 sizes (40, 80, 160, 320). Same with it? For example, jemalloc employs more fine-grained pools. ... -
- Thank you for your work. I have several questions.
* Could you explain the RString layout?
* Could you explain the variable size page strategy? https://bugs.ruby-lang.org/issues/18045 shows 4 sizes (40, 80, 160, 320). Same with it? F... -
- # GitHub PR: https://github.com/ruby/ruby/issues/4933
# Feature description
Since merging #18045 which introduced size pools inside the GC to allocate various sized slots, we've been working on expanding the usage of VWA into more ... -
- - [Feature #18239] Variable Width Allocation: Strings
- Discuss deployment schedule and binary incompatibility issues. -
- Solaris CI still has a problem even with these commits, so it doesn't
appear to fix the issue. Reverting both 84e8e2a39bba874433b661bd378165bd03c9d6aa
and bfd2f159f0c60ef8ac5bce6042edd25a571769b7. -
- Some comments:
* For debugger, `Thread::Backtrace::Location#script_lines` is enough.
* If we introduce this feature outside of `RubyVM`, people can use it casually and I'm not sure we can support it well because this feature (keeping... -
- I clarified a bit with @ko1.
First, `script_lines` returns all lines, not the lines of a specific method (I missed that in the description).
The debugger use case is to show the code around a breakpoint.
Adding something like `calle... -
- > For what else do you need to get the source? Ruby::AbstractSyntaxTree::Node?
> ...
This is why we propose this feature.
> Is it really needed on RubyVM::InstructionSequence?
> ...
I agree that.
-
- ko1 (Koichi Sasada) wrote in #note-13:
> Do you talking about `toSource` method in JS?
Yeah it's JS Function's `toSource`, but since it's an object it would also provide more information like line & column info.
> ...
For what els... -
- Eregon (Benoit Daloze) wrote in #note-12:
> We already have something similar to "code locations" objects: `Thread::Backtrace::Location`.
> ...
Do you talking about `toSource` method in JS?
> We can implement ISeq#source method like... -
- We already have something similar to "code locations" objects: `Thread::Backtrace::Location`.
We could add `#{first,last}_{column,line}` and `#code` to `Thread::Backtrace::Location`, how about that?
And then add a way to get a `Thread:... -
03:05 AM Feature #18231: `RubyVM.keep_script_lines`
- I don't want to bother progressing here, so I accept `RubyVM.keep_script_lines` with the indication of implementation dependency (and possible fragility).
When we got a better place, let's move on then.
Matz.
-
- For the `test-bundled-gems`, make `debug.so` with extconf.rb and
`make` command directly because `rake-compiler` assume ruby is
installed (but `test-bundled-gems` can run without installation). -
-
-
- Applied in changeset commit:git|b44c5187b4d05d8644242e532f2243210d0b74df.
----------
Expose instruction information for debuggers [Feature #18026] -
- Remove temporary skip of test_no_curdir to see if this fixes the
problem. -
- This appears to be only necessary on Solaris, but this commit
enables it unconditionally to test breakage. The following
commit will switch to only enabling it on Solaris. -
- 03:41 PM Revision aa2a1fb0 (git): * 2021-10-05 [ci skip]
-
-
- * Split the sorting types into classes.
* Apply the same sorting to method sorting under the parallel
test. -
-
- So that rubygems installer will work to build extension gems.
-
- Extension gems will be installed by the installed standard
libraries. -
- Could you open this on https://github.com/ruby/csv ? ruby/csv is the upstream of csv.
-
08:41 AM Bug #18238 (Third Party's Issue): CSV encoding issue with parsing from Zlib::GzipReader stream
- Hi,
I found an issue with parsing CSVs directly from a `Zlib::GzipReader` IO which I am trying to debug. Unfortunately, I am not at liberty to share the (proprietary) CSV file and I couldn't recreate the issue with a simplified/obfusc... -
-
-
-
- From recent results, `check` and `test-all` will finish within 10
minutes on macOS or 30 minutes on MinGW, otherwise time out. -
- Closing since this seems not acceptable.
Feel free to reopen if any progress. -
- https://github.com/ruby/irb/commit/692eb9b9b5
10/03/2021
-
-
11:15 PM Revision 3f5b52bf (git): Function `w32_error` does not return
-
- 03:14 PM Revision 76138316 (git): * 2021-10-04 [ci skip]
-
- https://github.com/ruby/reline/commit/852e855d82
-
-
01:16 PM Bug #18237 (Closed): Remove unnecessary checks for `Time` in `Range#each` as per the comment - Applied in changeset commit:git|fb03598666ddbb9f8b35eb4f599e75654ca7ca06.
----------
Remove unnecessary checks in `Range#each` [Bug #18237]
In commit:7817a438eb1803e7b3358f43bd1f38479badfbdc, the implementation
of `Time#succ`, which ha... -
11:50 AM Bug #18237 (Closed): Remove unnecessary checks for `Time` in `Range#each` as per the comment
- In commit:7817a438eb1803e7b3358f43bd1f38479badfbdc, the implementation of `Time#succ`, which had been deprecated for 10 years, was finally removed.
During that time, there was an explicit `instance_of?` check in source:range.c#L350 wi... - 01:16 PM Revision fb035986 (git): Remove unnecessary checks in `Range#each` [Bug #18237]
- In commit:7817a438eb1803e7b3358f43bd1f38479badfbdc, the implementation
of `Time#succ`, which had been deprecated for 10 years, was finally
removed.
During that time, there was an explicit `instance_of?` check in
source:range.c#L350 with... -
- Applied in changeset commit:git|19f9d9cf739e7fc185ef90d5da5b4b12cf902a52.
----------
Remove extraneous conversion to float [Bug #18236] -
-
- `rdoc` has been required so it looks unnecessary
https://github.com/ruby/irb/commit/5f749c613c -
- https://github.com/ruby/reline/commit/12ea63e6e0
-
- https://github.com/ruby/irb/commit/15bcbbb284
-
- https://github.com/ruby/irb/commit/800e83eabd
-
- https://github.com/ruby/reline/commit/1c6c0c82f0
-
- https://github.com/ruby/reline/commit/b42c0980d9
-
- I forgot to mention, it's also much better if all instances of a class (and potential subclasses) are immutable, if only part of the instances it's quickly confusing and most of the advantages disappear as the class is no longer truly im...
-
- I think nobody expects #freeze or #deep_freeze to ever freeze (non-singleton) classes/modules, so IMHO these methods simply not attempt that (except `SomeModule.freeze` of course).
It's the difference between state (ivars, and values of... -
- duerst (Martin Dürst) wrote in #note-8:
> Can you say exactly what the problem is? Is it that the file may get changed?
Yes, it might get changed, which might then cause confusing errors when parsing it again.
The file could also ha... -
- ko1 (Koichi Sasada) wrote in #note-6:
> > I would think having code_location on exceptions and methods is also enough for the debugger.
> ...
`code_location` would be an object, so e.g., `method(:foo).code_location.code` or `method(:fo... -
- Eregon (Benoit Daloze) wrote in #note-5:
> I am strongly against this proposal as it currently stands.
> ...
The `dead_end` gem (that we will try to integrate into Ruby early next year, see #18159) also needs code, and as far as I un... -
- https://github.com/ruby/reline/commit/1b00cae0fd
-
- Applied in changeset commit:git|ea64e742f5feddbdfb6526cd0a54a9986097a34d.
----------
Revert "mkmf.rb: try linking at try_var" [Bug #18235]
This reverts commit 524513be399e81bb170ec88aa0d501f33cbde8c3,
which can return false positive by... -
- This reverts commit 524513be399e81bb170ec88aa0d501f33cbde8c3,
which can return false positive by existing but unusable symbol,
including functions. -
- ruby_3_0 f192e01233ccd7966721a164e62a86707b4e6d9e merged revision(s) 7c0230b05d0978958f89434c84ddd9c82419c1a5,552728a23aeab0df598b356b19a573259e297d14,49af9012a20a824542cf588e55e5488895553e09.
-
- ruby_3_0 5341eca588e738cd5031ab7d8bb5c300471c00e9 merged revision(s) ddb32e66160ab50849419ef7c7ac584913b79c34.
-
- Check the entire name as `ruby2_keywords_flag` [Bug #18184]
---
marshal.c | 2 +-
test/ruby/test_marshal.rb | 8 +++++++-
2 files changed, 8 insertions(+), 2 deletions(-)
Check t... -
- [Bug #18173] Update loaded_features_index
If $LOADED_FEATURES is changed in the just required file, also the
index table needs to be updated before loaded_features_snapshot is
reset. If the snapshot is reset wit... -
- https://github.com/ruby/reline/commit/b283791dc5
-
- https://github.com/ruby/reline/commit/7fe9ecf945
-
- https://github.com/ruby/reline/commit/5414a42b0e
-
- If the argument prompt disappears when pasting is finished, rerender the
whole buffer.
https://github.com/ruby/reline/commit/996bbf8797 -
- https://github.com/ruby/reline/commit/65b27dd2ff
-
05:26 AM Revision ef350b3a (git): [ruby/reline] Simplify SIGWINCH handler to avoid aborting when resizing.
- https://github.com/ruby/reline/commit/481add0537
-
- https://github.com/ruby/reline/commit/2a60aacebb
-
- https://github.com/ruby/reline/commit/179c52bca8
-
- https://github.com/ruby/reline/commit/737b23beae
-
- https://github.com/ruby/reline/commit/da35902d78
-
- Because Ctrl+j and Enter are the same on Windows.
https://github.com/ruby/reline/commit/fa885ed449 -
- 02:23 AM Revision ceeae319 (git): Avoid using the altzone variable in AIX
- In AIX, altzone exists in the standard library but is not declared
in time.h. By 524513be399e81bb170ec88aa0d501f33cbde8c3, have_var
and try_var in mkmf recognizes a variable that exists in a library
even when it is not declared. As a r... - 02:09 AM Revision 2962c421 (git): [DOC] fix small mistake in doc/marshal.rdoc [ci skip]
-