DevMeeting-2024-06-06

https://bugs.ruby-lang.org/issues/20435

DateTime and location

2024/06/13 (Thu) 13:00-17:00 JST @ Online

Next Date

2024/07/11 (Thu) 13:00-17:00 JST @ Online

Announce

About release timeframe

Ordinary tickets

[Bug #20504] Interpolated string literal in regexp encoding handling (kddnewton)

I think there is an inconsistency here that we can solve by using the same codepath for creating regexps.
Could we change this? It should be minimal impact.

Preliminary discussion:

# encoding: us-ascii

interp = "\x80"
regexp = /#{interp}/
p regexp.encoding #=> #<Encoding:BINARY (ASCII-8BIT)>

regexp = /#{"\x80"}/ #=> invalid multibyte character: /\x80/

Discussion:

Kevin: One way or another this is inconsistent right now.
shyouhei: I guress nobody actually cared until now.
shyouhei: no new code should rely on this gotcha but is there any existing code?
Kevin: Is it possible for old codes to raise exceptions?
shyouhei: maybe not. The syntax error happens for literals, which means the source file becomes unloadable.
Kevin: agreed.

Conclusion:

matz: Let’s just be consistent. /#{"\x80"}/ should be a binary encoded regexp.

[Bug #20478] Circular parameter syntax error rules (kddnewton)

I think there is an inconsistency here, impacted by the presence of pipes.
Could we change this to be consistent? Either all reads at depth 0 or all reads at any depth.

Preliminary discussion:

# syntax error
def foo(bar = -> { bar }) end       # no lambda parameters
def foo(bar = ->() { bar }) end     # no lambda parameters
def foo(bar = baz { bar }) end      # no block parameters
def foo(bar = baz { _1 + bar }) end # parameters, but no pipes
def foo(bar = baz { it + bar }) end # parameters, but no pipes

# no syntax error
def foo(bar = ->(baz) { bar }) end   # lambda parameters
def foo(bar = baz { || bar }) end    # no block parameters but empty pipes
def foo(bar = baz { |qux| bar }) end # block parameters

def foo(bar = bar + 1) end
def foo(bar = bar) end

bar = bar + 1
#=> undefined method `+' for nil (NoMethodError)

bar = -> { bar.call }
bar.call #=> infinite loop

https://bugs.ruby-lang.org/issues/10314
mame: this warning was introduced just for compatibliity to very old behavior before 2.1.
mame: I guess no one was confused by accidentally writing def foo(x = x) or something
mame: I think now we can just remove this error

Discussion:

mame: I guess nobody would write def foo(bar=bar)
ko1: Actually we intentionally changed def foo(bar=bar) behaviour in 2.0 era. Before, bar=bar assigned a return value of method bar to a variable bar. cf: #10314
matz: I’d like to lean on syntax erros.
mame: Should def foo(bar = ->(baz) { bar }) also be an error? the bar can be visible.
matz: What should be the value of bar ? nil?
ko1: It would also make sense for them to just evaluate to nil. def foo(bar=bar) is already SyntaxError.
naruse: I don’t think bar = ->{bar.()} can be banned.
knu: there would be no way to make a recursive lambda then. :+1: for allowing this.
matz: not entirely persuadad but OK.
akr: why you prefered errors?
matz: I think earlier error should benefit. But agree with @knu that we need that kind of lambdas.

Conclusion:

matz: We don’t change ordinal ( non-argument ) local variables, at least.
matz: Let’s make def foo(bar=bar) assign nil to bar, even when there is a method namely bar.
matz: I accept the removal of all the warnings/exceptions

[Misc #20488] Document source file size restrictions (kddnewton)

Could we restrict source files to 32-bits?
It would help on memory size for parsers.

Preliminary discussion:

eval 'pp caller(0)', binding, 'hoge', (2 ** 31 - 1)
#=> ["hoge:2147483647:in `<main>'", "t.rb:2:in `eval'", "t.rb:2:in `<main>'"]

eval 'pp caller(0)', binding, 'hoge', (2 ** 31)
#=> win64: in `eval': bignum too big to convert into `long' (RangeError)
#=> linux: in 'Kernel#eval': integer 2147483648 too big to convert to 'int' (RangeError)

mame: currently, negative lineno is allowed

eval("raise", nil, "foo", -(2**31))
#=> foo:-2147483648:in '<top (required)>': unhandled exception

eval("raise", nil, "foo", -(2**31)-1)
#=> integer -2147483649 too small to convert to 'int' (RangeError)

mame: there is a dirty hack to eval("# some magic comment\n# and something\n" + src, nil, file, -1)
nobu: I have used the hack for some case

Discussion:

ko1: This is about cruby’s documentation, not ruby in general?
Kevin: yes but all rubies would eventually use it.
mame: In fact one could pass negative numbers to eval. The documentation should at least document about it.
shyouhei: I don’t remember where do we save this info. They were packed into NODE’s flags before but not any longer?
Kevin: It’s not stored anyware actually.

Conclusion:

matz: I agree with making the limits
- maximum line: signed 32 bits
- maximum column: unsigned 32 bits
- maximum file byte size: unsigned 32 bits

[Bug #20433] Hash.inspect for some hash returns syntax invalid representation (jeremyevans0)

Some symbols when used as hash keys results in Hash#inspect returning invalid syntax.
Most compatible fix is only using spaces around => when necessary.
Most consistent fix would be always using spaces around =>.
Alternative fix would be to have Symbol#inspect quote the symbols, even though quotes aren’t required outside of usage as a hash key.
Which approach should be used?

Preliminary discussion:

{ :a! => 1 } # {:a!=>1}
{ :a? => 1 } # {:a?=>1}
{ :== => 1 } # {:===>1}
{ :< => 1 }  # {:<=>1}
{ :+ => 1 }  # {:+=>1} # parseable!
{ :- => 1 }  # {:-=>1} # parseable!
{ :* => 1 }  # {:*=>1} # not parseable!
{ :/ => 1 }  # {:/=>1} # not parseable!
{ :/ => 1 }  # {:%=>1} # not parseable!
{ :< => 1 }  # {:<=>1}   # not parseable!
{ :<= => 1 } # {:<==>1}  # parseable!
{ :> => 1 }  # {:>=>1}   # not parseable!
{ :>= => 1 } # {:>==>1}  # parseable!
{ :<=> => 1} # {:<=>=>1} # parseable!

# Sol. 0
{ :== => 1 } # {:===>1} # no change

# Sol. 1
{ :== => 1 } # {:== =>1}
{ :a! => 1 } # {:a! =>1}
{ :key => 1 } # {:key=>1}

# Sol. 2
{ :== => 1 } # {:== => 1}
{ :a! => 1 } # {:a! => 1}
{ :key => 1 } # {:key => 1}

# Sol. 3
{ :== => 1 }  # {:"=="=>1}
{ :a! => 1 }  # {:"a!"=>1}
{ :key => 1 } # {:key=>1}

# Sol. 4
{ :== => 1, :key => 2 } # {"==": 1, key: 2}
{ :== => 1, :key => 2, "str" => 3 } # {"==": 1, key: 2, "str" => 3}

Discussion:

shyouhei: It’s very cryptic right now, at least.
mame: Jeremy proposes to add spaces (always / when needed).
matz: “when needed” is not as easy as it sounds.
matz: I don’t bother if it evaluates back. But user experience does matter.
knu: Solution #3 is a natural extension to the status quo. {:'foo bar' => 1}.inspect already quotes.
matz: I’d prefer Solution #2.
mame: would break tests though.
shyouhei: DO NOT PARSE #inspect OUTPUT. period.
mame: I’d push Solution #4 if we are allowed to break tests.

Conclusion:

~~mame: Let’s take solution #3. Quote symbols that match /\W\z/.~~
matz: Evaluate how solution #4 breaks existing tests. Can someone create a patch?

[Bug #20043] `defined?` checks for method existence but only sometimes (jeremyevans0)

I have a pull request to fix this (needs an update for prism).
However, @ko1 mentioned we may want to deprecate defined? when used with expressions.
Do we want to deprecate such usage of defined??

Preliminary discussion:

$ ./miniruby -e'p defined?(a)'
nil
$ ./miniruby -e'p defined?([a])'
nil
$ ./miniruby -e'p defined?([*a])'
"expression"

$ ./miniruby -e'p defined?(itself)'
"method"
$ ./miniruby -e'p defined?(itself(a))'
nil
$ ./miniruby -e'p defined?(itself(*a))'
"method"

$ ./miniruby -e'p defined?([[[[a]]]])'
nil
$ ./miniruby -e'p defined?({ a => a })'
"expression"

defined?(foo.bar.baz.qux) # foo.bar.baz is evaluated normally, and check if qux is defined on the object

defined?(foo)
defined?($foo)
defined?($1)
defined?(@@foo)
defined?(@foo)
defined?(Foo)
defined?(yield)
defined?(super)
defined?(foo.bar)

Discussion:

matz: all of them should return nil
akr: what’s defined supposed to do?
nobu: “evaluates right before the last method call”, implementation wise.
mame: would make sense to restrict what can be passed to it.
matz: I don’t think it’s worth breaking existing codes.

~2011 https://gist.github.com/ko1/b31517a5037d55bbe50e7f12d79b9fc1

Conclusion:

matz: I agree that defined?([*a]), defined?(itself(*a)), and defined?({ a => a }) should all return nil.
matz: I am not positive for removal of the currently-allowed expressions
ko1: I will investigate the use cases of defined? in gem codesearch

[Bug #19749] Confirm correct behaviour when attaching private method with `#define_method` (jeremyevans0)

I think the behavior of not changing method visibility is definitely a bug.
Last reviewed at the August 2023 developer meeting, still waiting on a decision from @matz.

Preliminary discussion:

class C
  def foo; end

  private

  define_method(:foo, instance_method(:foo))
  # current: the method foo is kept public
  # expected: the method foo is changed private
end

Conclusion:

matz: It is good to fix the issue. Accept the change

[Bug #18622] const_get still looks in Object, while lexical constant lookup no longer does (jeremyevans0)

It would be nice to have consistent behavior.
Last reviewed at the June 2023 developer meeting, still waiting on a decision from @matz.
- https://github.com/ruby/dev-meeting-log/blob/master/2023/DevMeeting-2023-06-08.md#bug-18622-const_get-still-looks-in-object-while-lexical-constant-lookup-no-longer-does-jeremyevans0
  - matz: I will reply

Preliminary discussion:

module ConstantSpecsTwo
  Foo = :cs_two_foo
end

module ConstantSpecs
end

p ConstantSpecs.const_get("ConstantSpecsTwo::Foo") # => :cs_two_foo
p ConstantSpecs::ConstantSpecsTwo::Foo # => const_get.rb:9:in `<main>': uninitialized constant ConstantSpecs::ConstantSpecsTwo (NameError)

Discussion:

shyouhei: const_get cannot be changed. We see there are use cases depend on this behaviour in the wild.

Conclusion:

matz: I’m just writing a reply right now… replied.

[Bug #20513] the feature of kwargs in index assignment has been removed without due consideration of utility, compatibility, consistency and logic (bughit)

it doesn’t make sense to remove a feature because there are fixable edge-case bugs in its implementation
it doesn’t make sense from a higher level language design perspective to allow kwargs in index but not in index assignment
instead of removal, index assignment should get proper/real kwargs and the multiple assignment problem should be fixed

Preliminary discussion:

https://bugs.ruby-lang.org/issues/20218

class Foo
  def []=(idx, opt={}, val)
  end
end

foo = Foo.new
foo[42, kw: "foo"] = :val

ary[kw:1] = 1
# keyword arg given in index assignment (SyntaxError)

https://gist.github.com/ko1/ed58eaef0cbcb552ce430ef9be3a7221

["gem_date/2009/11/23/amp-pure-0.5.0/lib/amp/server/repo_user_management.rb",
 (183,10)-(183,56),
 "REPOS[:url => repository] = {:private => true}"]
["gem_date/2009/03/12/colincasey-sequel-2.10.4/spec/sequel_core/dataset_spec.rb",
 (1665,4)-(1665,27),
 "@d[:a => 1] = {:x => 3}"]
["gem_date/2009/03/23/epugh-sequel-0.0.0/spec/core/dataset_spec.rb",
 (1668,4)-(1668,27),
 "@d[:a => 1] = {:x => 3}"]
{:total=>3}
2010
["gem_date/2010/04/22/viking-sequel-3.10.0/spec/core/dataset_spec.rb",
 (2071,4)-(2071,27),
 "@d[:a => 1] = {:x => 3}"]
["gem_date/2010/04/22/viking-sequel-3.10.0/spec/integration/dataset_test.rb",
 (539,4)-(539,26),
 "@ds[:a=>20] = {:b=>30}"]
["gem_date/2010/03/13/amp-0.5.3/lib/amp/server/repo_user_management.rb",
 (183,10)-(183,56),
 "REPOS[:url => repository] = {:private => true}"]
["gem_date/2010/11/29/rocket-server-0.0.4/lib/rocket/server/session.rb",
 (34,8)-(34,60),
 "subscriptions[channel => connection.signature] = sid"]
["gem_date/2010/11/02/bike-0.2.3/lib/_storage/sequel.rb",
 (158,8)-(158,47),
 "@dataset[:full_name => full_name] = val"]
["gem_date/2010/11/02/bike-0.2.3/lib/_storage/sequel.rb",
 (168,6)-(170,7),
 "@dataset[:full_name => v[:full_name]] = v.merge("]
{:total=>6}
2011
{}
2012
["gem_date/2012/09/25/ntable-0.1.6/test/tc_basic_values.rb",
 (108,8)-(108,51),
 "table_[:column => :blue, :row => 5] = \"bar\""]
["gem_date/2012/09/25/ntable-0.1.6/test/tc_basic_values.rb",
 (118,8)-(118,54),
 "table_[1 => 2, 0 => ::NTable.index(5)] = \"bar\""]
["gem_date/2012/09/25/ntable-0.1.6/test/tc_basic_values.rb",
 (140,8)-(140,43),
 "t1_[:row => 0, :column => :red] = 1"]
["gem_date/2012/09/25/ntable-0.1.6/test/tc_basic_values.rb",
 (142,8)-(142,43),
 "t2_[:row => 0, :column => :red] = 1"]
["gem_date/2012/06/14/rarity-0.1.1/lib/rarity/tracker.rb",
 (34,8)-(34,185),
 "@db[:images][:path=>cleanpath] = {:png_o_level => options[:png_o_level], :after_bytes=>results[:after], :time_taken=>(results[:time]+ @db[:images][:path=>cleanpath].time_taken)}"]
["gem_date/2012/12/27/fluent-plugin-couchbase-0.0.2/lib/fluent/plugin/out_couchbase.rb",
 (52,8)-(52,67),
 "connection[record.delete('key'), :ttl => self.ttl] = record"]
{:total=>6}
2013
2013
["gem_date/2013/04/22/hikkmemo-1.0.0/lib/hikkmemo/worker.rb",
 (33,16)-(33,65),
 "@db[:threads][:id => tid] = { :last_post => pid }"]
["gem_date/2013/10/20/multi_git-0.0.1/lib/multi_git.rb",
 (13,2)-(13,45),
 "BACKENDS[   :git, priority: 0] = GitBackend"]
["gem_date/2013/10/20/multi_git-0.0.1/lib/multi_git.rb",
 (14,2)-(14,48),
 "BACKENDS[:rugged, priority: 1] = RuggedBackend"]
["gem_date/2013/10/20/multi_git-0.0.1/lib/multi_git.rb",
 (15,2)-(15,46),
 "BACKENDS[  :jgit, priority: 2] = JGitBackend"]
["gem_date/2013/09/02/NetApp.rb-0.1.2/lib/netapp.rb",
 (168,16)-(172,17),
 "plexes[name: plx.child_get_string(\"name\")] = { "]
["gem_date/2013/09/02/NetApp.rb-0.1.2/lib/netapp.rb",
 (277,16)-(281,17),
 "plexes[name: plx.child_get_string(\"name\")] = { "]
["gem_date/2013/09/02/NetApp.rb-0.1.2/lib/netapp.rb",
 (471,12)-(473,13),
 "result[qtree: key.child_get_string(\"qtree\")] = {"]
["gem_date/2013/09/02/NetApp.rb-0.1.2/lib/netapp.rb",
 (484,12)-(491,13),
 "result[id: key.child_get_string(\"id\")] = {"]
["gem_date/2013/09/02/NetApp.rb-0.1.2/lib/netapp.rb",
 (549,12)-(555,13),
 "result[qtree: key.child_get_string(\"qtree\")] = {"]
["gem_date/2013/09/04/gentooboontoo-gphys-1.3.1.1/lib/numru/gphys/gphys.rb",
 (1508,3)-(1508,19),
 "gg['y'=>1] = 999"]
{:total=>10}
2014
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (19,6)-(19,39),
 "x[:x => 1 , :y => 1, :a => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (23,6)-(23,29),
 "x[:x => 1, :y => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (31,6)-(31,38),
 "x[:x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (32,6)-(32,38),
 "x[:x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (33,6)-(33,38),
 "x[:x => 1, :y => 1, :z => 3] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (35,6)-(35,38),
 "x[:x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (43,4)-(43,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (47,4)-(47,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (55,4)-(55,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (62,4)-(62,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (63,4)-(63,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (66,4)-(66,54),
 "y[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (67,4)-(67,54),
 "y[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (74,4)-(74,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (75,4)-(75,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (78,4)-(78,18),
 "y[:z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (79,4)-(79,18),
 "y[:z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (87,4)-(87,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (88,4)-(88,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (89,4)-(89,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 3] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (91,4)-(91,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (92,4)-(92,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (93,4)-(93,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 3] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (106,4)-(106,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (107,4)-(107,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 2"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (108,4)-(108,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 3] = 3"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (110,4)-(110,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 1] = 4"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (111,4)-(111,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 2] = 5"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (112,4)-(112,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 3] = 6"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (123,4)-(123,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 1] = 4"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (124,4)-(124,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 1, :z => 2] = 5"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (125,4)-(125,54),
 "x[:i => 1, :j => 1, :x => 1, :y => 2, :z => 3] = 6"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (135,4)-(135,27),
 "x[:x => 2, :y => 2] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (145,4)-(145,36),
 "x[:x => 1, :y => 1, :z => 1] = 1"]
["gem_date/2014/02/13/multimatrix-0.6/test/test_multimatrix.rb",
 (146,4)-(146,36),
 "x[:x => 2, :y => 1, :z => 1] = 1"]
{:total=>35}
2015
["gem_date/2015/05/15/olfactory-0.2.1/spec/olfactory/dictionary_spec.rb",
 (10,6)-(10,84),
 "Olfactory.dictionaries[:streets][:borough => 1, :street => \"BROADWAY\"] = 10001"]
2016
{}
2017
{}
2018
{}
2019
["gem_date/2019/03/19/gphys-1.5.6/lib/numru/gphys/gphys.rb",
 (1604,3)-(1604,19),
 "gg['y'=>1] = 999"]
{:total=>1}
2020
{}
2021
{}
2022
["gem_date/2022/05/19/simplygenius-atmos-0.14.0/lib/simplygenius/atmos/cli.rb",
 (193,12)-(193,51),
 "Atmos.config.[]=(k, v, additive: false)"]
{:total=>1}
2023
{}
2024
{}

Discussion:

mame: So we prohibited keyword arguments inside of arrays. The OP complains that they did use that feature.
matz: How did that behave?
mame: It was as if a hash was passed to the method.
ko1: I searched for older gems and it seems there were very little but actual usages.
ko1: Should we ignore them or not?
matz: Hmmmmm
mame: Honestly I don’t want people to use this though.

Conclusion:

matz: Let me communicate in #20218

[Feature #19998] Emit deprecation warnings when the old (non-Typed) Data_XXX API is used (byroot)

Marked as deprecated in the documentation as of Ruby 2.3.0 (2015)
Yet I found gems using it without realizing TypedData is preferred.
Proposed patch emit a deprecation warning when the gem is compiled: https://github.com/ruby/ruby/pull/10872

Discussion:

mame: It seems nobody is against issuing deprecation warnings.
shyouhei: Nobody hardly reads the compile-time output though. gem install doesn’t show them.
byroot: The intended audience is gem authors, they see warnings when they run bundle exec rake compile

Conclusion:

matz: go ahead

hsbt: I’m concerning that 3.4 has much more warnigns than before.

[Bug #20518] Escaped-newline in %W (akr)

I think it is a bug but fixing it causes (hopefully small) incompatibility.

Preliminary discussion:

'x\
foo\bar'   #=> current: "x\\\nfoo\\bar"

"x\
foo\bar"   #=> current: "xfoo\bar"

%w[x\
foo\bar]   #=> current: ["x\nfoo\\bar"]
%w[a\ b c] #=> current: ["a b", "c"]

%W[x\
foo\bar] #=> current: ["x\nfoo\bar"]
         #=> expected?: ["xfoo\bar"]
%W[x#{"\
"}foo\bar]

%I[x\
foo\bar] #=> [:"x\nfoo\bar"]

%q[x\
foo\bar] #=> "x\\\nfoo\\bar"

Discussion:

akr: I thought it was a bug at first sight. Not that sure right now though.
akr: newline in %w could either be a word separator, line separator, or a verbatim newline. Which one should this be?
matz: I know what is going on.
mame: I can live with the current situation.
akr: I now think we don’t have to risk breaking backwords compatibility for this one.

Conclusion:

akr: Am trying to improve documents.

[Misc #20407] Question about applying encoding modifier to an interpolated Regexp (andrykonchin)

The documentation states that a Regexp with only US-ASCII characters has US-ASCII encoding, otherwise a regular expression is assumed to use the source encoding. This can be overridden with encoding modifiers.
It isn’t clear enough how a Regexp encoding (that Regexp#encoding method returns) is calculated in case an encoding modifier (e.g. u, e, etc) is specified for a Regexp with interpolation (e.g. /a #{ "b".encode("windows-1251") } c/e).
The encoding modifier is applied in some cases and isn’t in other ones what seems confusing at first glance.
It seems result depends on a source encoding, characters of the Regexp literal fragments and encoding of Regexp interpolated fragments as well.

Discussion:

nobu: /.../e is merely a hint. The contents of the regular expression should be honoured.
mame: It’s very hard for me to consider //e as a hint. It doesn’t make sense for instance //i to be ignored.
akr: We would better remove //e.
mame: If we are going to remove should we try hard describing what is going on here right now?

Conclusion:

nobu: Consistent or not the current situation is that “if a regular expression, after interpolated, solely consist of US-ASCII characters, then it honours the encoding suffix. Otherwise the contents wins”.
akr: Deprecation of //e shall be a separate topic.

[Bug #20416] IO#read doesn’t preserve buffer encoding if `maxlen = nil` (andrykonchin)

IO#read and similar methods when called with buffer argument preserve its encoding.
But IO#read doesn’t do so in case the maxlen argument is nil.

Discussion:

nobu: I have commented to the ticket.

Conclusion:

matz: agreed with nobu. Closing

[Bug #20319] Singleton class is being frozen lazily in some cases (andrykonchin)

When an object becomes frozen (with #freeze method call) only its own singleton class becomes frozen immediately.
Classes in the singleton classes chain become frozen only when they are returned to user with #singleton_class method call.
The object’s singleton class’ singleton class (and so on) doesn’t become frozen even if it was already instantiated
This lazy freezing can be observed by a user when he gets a singleton class of the object’s singleton class before freezing the object. After freezing the object the instantiated singleton class is still not frozen.
There might be several options (1) don’t change anything, 2) freeze all the instantiated singleton classes in a chain immediately or 3) don’t freeze singleton classes in a chain at all and stop freezing even an object’s singleton class) and there is a PR with fix (https://github.com/ruby/ruby/pull/10245) but it’s unclear what option is the best one.

Discussion:

mame: Should a singleton class’ singleton class shall also be frozen when the object gets frozen?
shyouhei: An object’s singleton class also gets frozen when the object gets frozen is a twist.
mame: But there seems no way other than that to prevent adding a singleton method to a frozen object.
matz: Is it difficult to recursively freeze everything?
nobu: Not really. Jeremy already has a patch to do so.
(reviewing Jeremy’s patch)
mame: What is “shared singleton class”? The patch follows the attached_object chain to make sure it matches the original object. Why is this process needed?
shyouhei: I think there was a bug in the past where the singleton class was shared, but I don’t know if that had anything to do with it.
mame: According to Jeremy’s comment, it appears that a naive fix broke TestModule#test_frozen_singleton_class failed and he found this hack
mame: Looks like no one can understand this patch immediately. It would be difficult to review the patch in this meeting.

Conclusion:

matz: We cannot determine anything without deeper understanding

[Feature #6648] Provide a standard API for retrieving all command-line flags passed to Ruby (eregon)

@nobu are you OK with RbConfig.ruby_args? If not can you suggest a better place?

Discussion:

mame: What is needed is currently unable to obtain from ::ARGV.
matz: I’m OK with the feature.
nobu: I’m against adding it to RbConfig.
mame: @eregon is against adding it to RubyVM
nobu: Maybe Process?

mame: is it possible to execute the current ruby in Windows?
nobu: No.
shyouhei: There is no way to exec in Windows?
nobu: Yes there is but there is no way to reconstruct C level argv from what we currently have.
shyouhei: It’s NG then. What is impossible is impossible.
mame: I don’t say it is impossible, but I strongly feel a bad smell for this API

Conclusion:

mame: I will reply

[Feature #20443] Allow Major GC’s to be disabled

@ko1 (Koichi Sasada) is against the name disable_major
I would like advice from committers about what an appropriate name for this method should be.

Discussion:

matz: I think I would like to have this feature.
ko1: GC.disable_major or GC.disable_major_gc?
ko1: GC.needs_major? or GC.needs_major_gc? or GC.need_major_gc?
ko1: “major GC” is a GC terminology. I prefer “disable_major_gc”.
matz: what we currently do with what we call the “major GC” highly depends on our current GC implementation. Having this method could make other pluggable algorithms harder.
byroot: the plan is for plugable GC to declare what options they support. If they don’t support one method like this, it behave as NotImplemented.
- matz: OK then.
shyouhei: What happens when there are more than 3 generations is not clear to me.
Gc.major does a major GC: https://ocaml.org/manual/5.2/api/Gc.html
byroot: GC.start calls it full_mark, so it’s also an option I guess
nobu: GC.disable_full_mark or GC.disable(full_mark: true) or GC.disable(full_mark: false)
- ko1: true? false? keyword argument is too confusing
- byroot: keyword arguments are hard to feature test. e.g. GC.respond_to?(:disable_full_mark) is easy to use in gems etc.
  - mame: makes sense
matz: I prefer GC.disable_major and GC.need_major?(no s of needs)
byroot/ko1: How about more generalized setting mechanism?

GC.configure(
  full_mark: false,
  mmtk_something: 42,
  mmtk_other_thing: 4422,
)

# for the current GC impl
p GC.get_config #=> { enabled: true, full_mark: false, stress: false, auto_compact: true, ... }

# for the mmtk-based GC
p GC.get_config #=> { mmtk_something: 42, mmtk_other_thing: 4422, ...}

byroot: I like this because it helps plugable GC, but what would be the behavior on unknown keys? Ignored or error?
- nobu: error?
- byroot: if error, then we need a way to know what keys are accepted, otherwise it makes supporting multipe Ruby versions hard.
byroot: Also I think GC.config and GC.config = {} would feel a bit more natural?
byroot: Feature testing can be done with GC.config.key?(:full_mark)
mame: it is difficult to decide the details of the setting mechanism today. We should decide which approach (individual method like GC.disable_major or grand setting API) is preferred
matz: I prefer the generalized one
nobu: me too
ko1: me too
byroot: how to we handle GC.need_major? with the generalized solution? GC.config[:need_major]? Or do we keep GC.need_major? as a method.
- mame: good point
ko1: GC.latest_gc_info can contain information
- byroot: good idea.
- GC.latest_gc_info[:need_major_by]
ko1: we can extend GC.latest_gc_info(:need_major_by) like GC.stat(key)
- nobu: it is already supported.
- ko1: it can lack the information so we may need to introduce new key.

Conclusion:

matz: GC.configure-like API should be explored

[Bug #20314] Simultaneous Timeout expires may raise an exception after the block (mame)

I need a wisdom from committers on how to fix nested timeouts.

Preliminary discussion:

def task1
  Timeout.timeout(0.5, B) do
    nil while true
  rescue B
    log("timeout")
  end
rescue B
  log("timeout/outer")
ensure
  final1
  final2
end

def task2
  Timeout.timeout(0.1, A) do
    task1
    task3
  end
rescue Timeout::Error
end

Timeout.timeout(0.1, A) do # A
  Timeout.timeout(0.5, B) do # B
    # A のタイムアウトを一時停止して、B を 0.1 秒待つ（A の待ち時間）
    # stop_outer_timeout
    # timeout = outer < innner ? outer : innner
    sleep 1 # 0.1 秒後に B が投げられる
  ensure
    # ここに到達した時点で A を再開する restart_outer_timeout
    # が、0.1 秒たっているので、A を raise する
  end
ensure
end

Timeout.timeout(0.5, A) do # A
  Timeout.timeout(0.1, B) do # B
    # A を一時停止
    sleep 1 # 0.1 秒後に B が投げられる
  rescue B
    sleep 1 # これを Aで止めたいけど、A は一時停止しているので止められない
  ensure
    # ここに到達した時点で A を再開する restart_outer_timeout
    # が、0.1 秒たっているので、A を raise する
  end
ensure
end

Conclusion:

no conclusion. timeout API is difficult

DevMeeting-2024-06-06

DateTime and location

Next Date

Announce

About release timeframe

Ordinary tickets

[Bug #20504] Interpolated string literal in regexp encoding handling (kddnewton)

[Bug #20478] Circular parameter syntax error rules (kddnewton)

[Misc #20488] Document source file size restrictions (kddnewton)

[Bug #20433] Hash.inspect for some hash returns syntax invalid representation (jeremyevans0)

[Bug #20043] defined? checks for method existence but only sometimes (jeremyevans0)

[Bug #19749] Confirm correct behaviour when attaching private method with #define_method (jeremyevans0)

[Bug #18622] const_get still looks in Object, while lexical constant lookup no longer does (jeremyevans0)

[Bug #20513] the feature of kwargs in index assignment has been removed without due consideration of utility, compatibility, consistency and logic (bughit)

[Feature #19998] Emit deprecation warnings when the old (non-Typed) Data_XXX API is used (byroot)

[Bug #20518] Escaped-newline in %W (akr)

[Misc #20407] Question about applying encoding modifier to an interpolated Regexp (andrykonchin)

[Bug #20416] IO#read doesn’t preserve buffer encoding if maxlen = nil (andrykonchin)

[Bug #20319] Singleton class is being frozen lazily in some cases (andrykonchin)

[Feature #6648] Provide a standard API for retrieving all command-line flags passed to Ruby (eregon)

[Feature #20443] Allow Major GC’s to be disabled

[Bug #20314] Simultaneous Timeout expires may raise an exception after the block (mame)

[Bug #20043] `defined?` checks for method existence but only sometimes (jeremyevans0)

[Bug #19749] Confirm correct behaviour when attaching private method with `#define_method` (jeremyevans0)

[Bug #20416] IO#read doesn’t preserve buffer encoding if `maxlen = nil` (andrykonchin)