DevMeeting-2024-12-12

https://bugs.ruby-lang.org/issues/20879

DateTime and location

2024/12/12 (Thu) 13:00-17:00 JST @ Online

Next Date

2025/01/09 (Thu) 13:00-17:00 JST @ Online

Announce

About release timeframe

Ordinary tickets

[Bug #20930] Different semantics for nested `it` and `_1` (eregon)

I allowed multiple uses of it in different levels of nested blocks, e.g. files.each { YAML.parse_file(it).each { p it } }, which is a SyntaxError with _1. Can we confirm that it’s okay to introduce the behavior?

Discussion:

k0kubun: _1 is error when nested. But it has been around before _1. Was unable to prohibit this behaviour so allowed.
k0kubun: Eregon points out the inconsistency but I honestly want current behaviour. Thoughts?
mame: I don’t understand why _1 is prohibited to coexist.
nobu: because it’s too difficult.

Conclusion:

matz: keep the current behaviour.

[Feature #20878] A new C API to create a String by adopting a pointer (byroot)

Would be useful for many extensions.
Allow to efficiently work with a raw, malloced buffer, and then turn it into a Ruby string without copying it.
The pointer is adopted as soon as the function is called, if the function fails the pointer is freed immediately.
Proposed implementation: https://github.com/ruby/ruby/pull/12143
Current name is rb_enc_str_adopt, but rb_enc_str_move could be make sense too.

Discussion:

matz: I’m not in favour of the name.
shyouhei: I am, though. “Adopt” sounds the most right terminology.
nobu: I don’t think it’s necessary.
shyouhei: by which alternative?
nobu: create a string first and extract its pointer.
shyouhei: don’t extract a string pointer.
nobu: aaand it’s dangerous to adopt a bare pointer.
shyouhei: while I agree with that byroot says he makes sure it’s safe.
mame: I wonder if it’s safe to assume xfree can deallocate asprintf return value, once mmtk lands.
akr: I once thought it would be nice if we could adopt realpath(3) return value.

Conclusion:

no conclusion, but let us dump our insights.

[Feature #20861] Add an environment variable so we can change the thread quantum (tenderlovemaking)

It maintains the default quantum but allows applications to experiment with different quantums
Lower quantums can help improve performance for applications with mixed IO / CPU workloads
The best quantum depends on the application load, so having a configuration option would be nice
We discussed at RubyConf and the feature seems fine but the name maybe not good enough. What is an appropriate name?

Discussion:

ko1: per-thread quantum could not be achieved on some platforms (notably linux). there are technical difficulties.
matz: naming candidcates?
ko1: environment variables are legit here.

ko1:

RUBY_THREAD_DEFAULT_QUANTUM_MS
RUBY_THREAD_SWITCH_INTERVAL
RUBY_THREAD_TIMESLICE
RUBY_THREAD_TIME_QUANTUM

matz: is “quantum” a normal terminology in multithreaded studies?
ko1: not a strange naming I guess. Others are okay too.
nobu: is it in milliseconds?
ko1: yes. Makes no sense for it to be in both nanoseconds, and/or seconds. Msec is natural.
akr: does it take effect when a ruby script changes this environment variable?
ko1: Not right now. Do we need to?
nobu: doable though.
akr: ENV[“TZ”] does that. I don’t say we should also here, though.

Conclusion:

matz: RUBY_THREAD_TIMESLICE accepted.

[Feature #20884] reserve “Ruby” toplevel module for Ruby language (hsbt)

It’s better to reserve this at Ruby 3.4, not 3.5.

Discussion:

ko1: What happens to current RUBY_VERSION etc.?
nobu: Won’t be deleted today.
hsbt: Today? I guess forever.
matz: +1 for domain squatting, concerning this leads lots of bikeshed pull requests…
mame: I have class Ruby already so kind of sad to me if module Ruby gets predefined, but I am okay for the change

ko1@ruby-sp1:~$ ~/gem-codesearch/gem-codesearch '^Ruby ='
2016-06-22 /srv/gems/goodguide-gibbon-0.14.4/vendor/gibbon/gibbon.browser.dev.js:Ruby = (function(_super) {
2016-06-22 /srv/gems/goodguide-gibbon-0.14.4/vendor/gibbon/gibbon.browser.js:Ruby = (function(_super) {
2011-02-15 /srv/gems/mbailey-chef-0.9.12.2/lib/chef/knife/bootstrap/client-install.vbs:Ruby = Dst & "\bin\ruby.exe"
2011-05-06 /srv/gems/naked-chef-0.9.15.3/lib/chef/knife/bootstrap/client-install.vbs:Ruby = Dst & "\bin\ruby.exe"
2009-07-04 /srv/gems/ripper2ruby-0.0.2/test/fixtures/stuff.rb:Ruby = 'Just another Ruby hacker,' and print Object.const_get(:Ruby)
2016-12-29 /srv/gems/ruby-compiler-0.1.1/vendor/ruby/tool/eval.rb:Ruby = ENV['RUBY'] || RbConfig.ruby
2016-05-20 /srv/gems/rubylexer-0.8.0/test/data/jarh.rb:Ruby = 'Just another Ruby hacker,' and print Object.const_get(:Ruby)
2009-07-04 /srv/gems/svenfuchs-ripper2ruby-0.0.2/test/fixtures/stuff.rb:Ruby = 'Just another Ruby hacker,' and print Object.const_get(:Ruby)

$ ~/gem-codesearch/gem-codesearch '^class Ruby\b'
2013-04-18 /srv/gems/_vm-0.0.0/_ruby:class Ruby < Manager
2022-05-29 /srv/gems/arachni-1.6.1.3/components/fingerprinters/languages/ruby.rb:class Ruby < Platform::Fingerprinter
2016-04-29 /srv/gems/core_extended-0.0.10/lib/core_extended/ruby.rb:class Ruby
2023-10-28 /srv/gems/cvefixer-0.6.8/lib/taskgroups/ruby.rb:class Ruby
2018-05-17 /srv/gems/dev_osx_updater-0.0.6/lib/ruby.rb:class Ruby
2011-05-20 /srv/gems/gdb.rb-0.1.7/scripts/ruby-gdb.py:class Ruby (gdb.Command):
2013-11-25 /srv/gems/gitlab-pygments.rb-0.5.4/vendor/pygments-main/tests/examplefiles/example.rb:class Ruby < Scanner
2008-09-17 /srv/gems/grammar-0.8.1/lib/grammar/ruby/code.rb:class Ruby
2008-09-17 /srv/gems/grammar-0.8.1/lib/grammar/ruby.rb:class Ruby
2008-09-17 /srv/gems/grammar-0.8.1/test/test_ruby.rb:class Ruby < ::Test::Unit::TestCase
2017-04-28 /srv/gems/inwxupdate-0.4.0/lib/detectors/ruby.rb:class Ruby
2011-09-10 /srv/gems/linqr-0.1.0/lib/linqr.rb:class Ruby::Node
2011-09-10 /srv/gems/linqr-0.1.0/lib/source_name_evaluator.rb:class Ruby::Arg
2011-09-10 /srv/gems/linqr-0.1.0/lib/source_name_evaluator.rb:class Ruby::Call
2011-09-10 /srv/gems/linqr-0.1.0/lib/source_name_evaluator.rb:class Ruby::Variable
2011-09-10 /srv/gems/linqr-0.1.0/lib/source_name_evaluator.rb:class Ruby::Const
2014-03-14 /srv/gems/mortar-pygments.rb-0.5.7/vendor/pygments-main/tests/examplefiles/example.rb:class Ruby < Scanner
2013-12-04 /srv/gems/pygments.rb-jruby-0.5.4.2/vendor/pygments-main/tests/examplefiles/example.rb:class Ruby < Scanner
2016-01-27 /srv/gems/test_mysore-0.0.0/lib/test_mysore.rb:class Ruby
2012-09-23 /srv/gems/xiki-0.6.5/lib/xiki/ruby.rb:class Ruby

nobu: for instance AST should not be there.
mame: in which version? -> 3.4
shyouhei: should we freeze the constant so that user land can never overwrite?
akr: maybe tests etc. want to mock these constants. I don’t think it’s worth.

Conclusion:

matz: global toplevel Ruby accepted, while what should be inside of it need very careful considerations for each one of them.
matz: Let’s warn first.
- 3.4: issue deprecation warning for toplevel constant ::Ruby (which ever class this is)
- 3.5: define the module.

[Misc #20913] Proposal: Adding Jeremy Evans and Burdette Lamar to `www.ruby-lang.org`’s English Editorial Team (st0012)

They have expressed enthusiasm for improving the website during recent discussions at RubyConf in Chicago.
Adding them would enhance our reviewing capacity for the increased contributions to the site’s content.

Discussion:

matz: Are both of them okay with this?
shyouhei: Seems @BurdetteLamar have not contributed to ruby-lang.org so far?
matz: I met him at Ruby Conf. Also he has contributed to ruby’s main repo.
hsbt: person aside, I want to know what RubyCentral talked with them. Do they plan to do anything on the site?

Conclusion:

Propsal itself is kind of welcome, but it would be better if the surrounding context is a bit detailed.

[Feature #20935] API for Globally Enabling/Disabling Happy Eyeballs Version 2 in the Socket Class (shioimm)

I would like to discuss the appropriateness of the API for enabling/disabling HEv2 in the Socket class.

Discussion:

shioimm: This feature itself would be necessary I guess. I would like to know if this API design is okay?
ko1: Tell us the use case of this option?
shioimm: There are cases when infrastructure is already known to have exact one DNS server. HEv2 would degrade performance for those cases and users might want to disable this feature.
matz: naming-wise, is fast_fallback acceptable here?
mame: I guess golang uses that terminology.
shyouhei: Is this method argument only? or environment variable?
mame: or both. would be nice to have them at once.
ko1: should we introduce an environment variable?
matz: I prefer envval.
ko1: RUBY_TCP_FAST_FALLBACK=0 for disable?
mame: RUBY_SOCKET_TCP_FAST_FALLBACK=0 to make clear that it is for Socket library?
matz: default (0) should enable happy eyeballs, so RUBY_TCP_NO_FAST_FALLBACK=1 disables happy eyeballs.
naruse: RUBY_TCP_FAST_FALLBACK_MODE=tradtional|happy_eyeballs_v2
https://github.com/ruby/ruby/blob/6a1aaf3679ed6d467172debc10de4df1c484a21b/man/ruby.1#L703

Conclusion:

matz: methods are okay.
matz: envval: RUBY_TCP_NO_FAST_FALLBACK=1 to disable happy eyeballs.

[Bug #20940] Colored syntax error from prism

just confirmation.
off-topic: RUBY_DESCRIPTION contains +PRISM?

Discussion:

matz: it is okay if we can disable coloring.
naruse: gcc/clang and other softwares support coloring.
akr: how about to support NO_COLOR?

$ ruby -e "foo(" |& cat
-e: -e:1: syntax error found (SyntaxError)
> 1 | foo(
    |     ^ unexpected end-of-input; expected a `)` to close the arguments
  2 |

matz: NO_COLOR=1 to disable coloring.

Conclusion:

matz: NO_COLOR=1 to disable coloring. 1 can be replaced with any non-empty string.
- NO_COLOR=0 should also disable color

$ ruby -v
ruby 3.4.0dev (2024-12-10T23:16:24Z master c71f7faaa9) +PRISM [x86_64-linux]
$ ruby -v --parser=parse.y
ruby 3.4.0dev (2024-12-10T23:16:24Z master c71f7faaa9) [x86_64-linux]

matz: remove +PRISM towards 3.5

[Feature #20875]: Atomic initialization for Ractor local storage

Discussion:

ko1: compute_if_absent terminology is introduced.
matz: compute is not known in Ruby world. Ractor.store_if_absent(key){ init_block } is aceptable. I prefer than init/once.
ko1Ractor.local_storage_store_if_absent(key){ init_block } (local_storage) is not needed?
matz: too long.
mame: atomicity (thread-safety) is not represented with this name. Is it okay?
matz: no problem.

Conclusion:

matz: Accept Ractor.store_if_absent(key){ init_block }

[#20858] please reject > matz

[#20817] Ruby 3.4.0dev emits `warning: possibly useless use of + in void context` while Ruby 3.3.5 does not

akr: we should accept such a minor behavior change, regardless of prism
mame: I will close

[#20757] Make rb_tracearg_(parameters|eval_script|instruction_sequence) public C-API

[#20882] Provide `Boolean(...)`

# ENV["SOME_FEATURE"] is unset
Boolean(ENV["SOME_FEATURE"]) # => false

# ENV["SOME_FEATURE"] is unset, but allow a default?
Boolean(ENV["SOME_FEATURE"], true) # => true

# explicitly disable
ENV["SOME_FEATURE"] = "0"
Boolean(ENV["SOME_FEATURE"], true) # => false

# explicitly enable
ENV["SOME_FEATURE"] = "1"
Boolean(ENV["SOME_FEATURE"]) # => true

matz: rejected

[#20885] `String#gsub?`

str.gsub?(...) == str.dup.gsub!(...)

# akr's proposal
str = "foo".freeze
str2 = str.gsub(/bar/, "BAR")
str.equal?(str2) #=> current: false, akr-proposed: true

# akr's proposal 2
str = "foo".freeze
# A new method "gs" returns self if no match
str2 = str.gs(/bar/, "BAR") # or sg
str2.frozen? #=> true
str.equal?(str2)

https://blade.ruby-lang.org/ruby-dev/33553

matz: I will reject

[#20899] Reconsider adding `Array#find_map`

(1..9).find_map {|i| i * 2 if i.even? } #=> 4

ret = (1..9).find {|i| i.even? } #=> 2 or nil
ret = ret * 2 if ret

ret = (1..9).find {|i| break i * 2 if i.even? }

ret = (1..9).find {|i| i.even? }&.then { it * 2 }

(1..9).find{it.even?}&.then{nil}
(1..9).find_then{nil if it.even?}

ret = (1..9).find do |i|
  v = f(i)
  cond?(v)
end&.then { f(ret) }

ret = (1..9).lazy.map {|i| f(i) }.find {|v| v }
ret = (1..9).find_map {|i| f(i) }
ret = (1..9).lazy_map_find {|i| f(i) }

ret = nil
(1..9).find {|i| ret = f(i) }

ret = nil
(1..9).each {|i| break if ret = f(i) }

ret = (1..9).map_find { f(it) }
ret = (1..9).map{ f(it) }.find{ it }

(1..9).find {|v| v.price >= 200 }&.price

list = ['some 123', 'list 234', 'of 345', 'strings 456']
list.find_map {|s| s[/\Aof (\d+)\z/, 1] } # => "345"
list.find_map {|s| $1.to_i if s =~ /\Aof (\d+)\z/ } # => 345
list.find {|s| break $1 if s =~ /\Aof (\d+)\z/ } # => "345"

# to not explicitly use word "map" if it feels confusing
find_mapped
find_transform
find_transformed

# to connect with `Object#then` method
find_then
find_and_then

# to explicitly connect to filter_map
filter_map_first

filter_map(first: true)

matz: I don’t understand the need very much, but the name find_map is acceptable. I will reply
nobu: Should we try it in Ruby 3.5, not 3.4?

[#13820] Add a nil coalescing operator

??: ATS, C#, JavaScript, PHP, PowerShell, Swift
No operator: Python, Ruby, Rust, SQL
?:: CFML, Kotlin, Objective-C
!: Freemarker, Haskell
:-: Bourne-like shells
|?: F#
//: Perl
%||%: R
If: VB.NET

def cached_foo(...)
  @cache ||= foo(...)

  @cache |||= foo(...)

  if @cache == nil
    @cache = foo(...)
  else
    @cache
  end
end

[#20925] Allow boolean operators at beginning of line to continue previous line

if cond1
  && cond2
  && cond3
  statement1
  statement2
end

if a = b && cond #=> a = (b && cond)
end

if a = b
  && cond
end

if request.secret_key_base.present? &&
  request.encrypted_signed_cookie_salt.present? &&
  request.encrypted_cookie_salt.present?

  request.encrypted_cookie
end

if cond1 & # concat & and &
 & cond2
  BODY
end

[#20922] Should not we omit parentheses in assert calls?

p -1    # user may expect "p - 1"
p -1, 2 # user cannot expect "p - 1, 2", no misleading

assert_equal -1, n         # ambiguous first argument; put parentheses or a space even after `-` operator
assert_equal {}, something # syntax error

assert_equal (-1), minus_one
assert_match (/foo/), str
assert_match $ /foo/, str

/foo/ |> assert_match _, str

re = /foo/
assert_match re, str

assert_match \
  /foo/, str

assert_match  /foo/, str

[#17566] Tune thread QoS / efficiency on macOS

[#20917] `redo`/`next` in nested `begin` block causes wrong order of execution

[#20938] Percent String literal delimiter impacts string contents with parse.y

[#11177] DATAでEOF文字以降が読めない

[#19191] Implicit console input transcoding is more desirable

[#20889] `IO#ungetc` and `IO#ungetbyte` should not cause `IO#pos` to report an inaccurate position

[#20919] `IO#seek` and `IO#pos=` do not clear the character buffer in some cases while transcoding

~https://bugs.ruby-lang.org/issues/20924 IO#readline ignores the limit argument when the encoding is UTF-32LE and the limit would split a character~

[#20943] Constant defined in `Data.define` block

module (expr)::Foo
  # cref??
end

module M
  B = 1
end
M.class_eval do
  p B # error
  A = 1
end
p A

module M1
  module Foo
    X = 1
  end
end
module M2
  module Foo
    X = 2
  end
end
[M1, M2].each do
  module it::Foo
    p [:A, X]
    if rand < 0.5
      p [:B, X]
    else
      p [:C, X]
    end
  end
end

# [:A, 1]
# [:C, 1]
# [:A, 1]
# [:B, 2]

module M1
  module Foo
    X = 1
  end
end
module M2
  module Foo
    X = 2
  end
end
[M1, M2].each do |m|
  module m::Foo
    def self.f
      p X
    end
  end
end

M1::Foo.f #=> 1
M2::Foo.f #=> 1

DevMeeting-2024-12-12

DateTime and location

Next Date

Announce

About release timeframe

Ordinary tickets

[Bug #20930] Different semantics for nested it and _1 (eregon)

[Feature #20878] A new C API to create a String by adopting a pointer (byroot)

[Feature #20861] Add an environment variable so we can change the thread quantum (tenderlovemaking)

[Feature #20884] reserve “Ruby” toplevel module for Ruby language (hsbt)

[Misc #20913] Proposal: Adding Jeremy Evans and Burdette Lamar to www.ruby-lang.org’s English Editorial Team (st0012)

[Feature #20935] API for Globally Enabling/Disabling Happy Eyeballs Version 2 in the Socket Class (shioimm)

[Bug #20940] Colored syntax error from prism

[Feature #20875]: Atomic initialization for Ractor local storage

[#20858] please reject > matz

[#20817] Ruby 3.4.0dev emits warning: possibly useless use of + in void context while Ruby 3.3.5 does not

[#20757] Make rb_tracearg_(parameters|eval_script|instruction_sequence) public C-API

[#20882] Provide Boolean(...)

[#20885] String#gsub?

[#20899] Reconsider adding Array#find_map

[#13820] Add a nil coalescing operator

[#20925] Allow boolean operators at beginning of line to continue previous line

[#20922] Should not we omit parentheses in assert calls?

[#17566] Tune thread QoS / efficiency on macOS

[#20917] redo/next in nested begin block causes wrong order of execution

[#20938] Percent String literal delimiter impacts string contents with parse.y

[#11177] DATAでEOF文字以降が読めない

[#19191] Implicit console input transcoding is more desirable

[#20889] IO#ungetc and IO#ungetbyte should not cause IO#pos to report an inaccurate position

[#20919] IO#seek and IO#pos= do not clear the character buffer in some cases while transcoding

[#20943] Constant defined in Data.define block

[Bug #20930] Different semantics for nested `it` and `_1` (eregon)

[Misc #20913] Proposal: Adding Jeremy Evans and Burdette Lamar to `www.ruby-lang.org`’s English Editorial Team (st0012)

[#20817] Ruby 3.4.0dev emits `warning: possibly useless use of + in void context` while Ruby 3.3.5 does not

[#20882] Provide `Boolean(...)`

[#20885] `String#gsub?`

[#20899] Reconsider adding `Array#find_map`

[#20917] `redo`/`next` in nested `begin` block causes wrong order of execution

[#20889] `IO#ungetc` and `IO#ungetbyte` should not cause `IO#pos` to report an inaccurate position

[#20919] `IO#seek` and `IO#pos=` do not clear the character buffer in some cases while transcoding

[#20943] Constant defined in `Data.define` block