Since Git 2.46, commands like `git commit` spawn a detached
`git maintenance run --auto` process, which briefly creates
`.git/objects/maintenance.lock`. Specs copying a just-committed
repository with FileUtils.cp_r can hit the lock file disappearing
between readdir and lstat, causing flaky Errno::ENOENT failures on
macOS CI runners with newer Git.
https://github.com/ruby/ruby/actions/runs/28764006107/job/85284933867
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
In #225 it was reported that the output looks incorrect:
```
$ cat /tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb
def x.y.z
end
$ ruby /tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb
/tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb: --> /tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb
expected a delimiter to close the parametersunexpected '.', ignoring it
> 1 def x.y.z
> 2 end
```
Specifically:
```
expected a delimiter to close the parametersunexpected '.', ignoring it
```
However this does not show up when executing the debug executable:
```
$ bin/bundle exec exe/syntax_suggest /tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb
--> /tmp/4a71c7e417cc9eac0971e3a2519b295c/scratch.rb
expected a delimiter to close the parameters
unexpected '.', ignoring it
> 1 def x.y.z
> 2 end
```
This is because `exe/syntax_suggest` uses STDOUT.puts while calling `ruby` with the filename uses a fake IO object represented by MiniStringIO. This class was incorrectly not adding a newline to the end of the print.
The fix was to move the class to it's own file where it can be tested and then fix the behavior.
close https://github.com/ruby/syntax_suggest/pull/225https://github.com/ruby/syntax_suggest/commit/d2ecd94a3b
Co-authored-by: Andy Yong <andyywz@gmail.com>
Method#unbind clones the method entry, preserving its defined_class.
For methods mixed in via include/extend, defined_class is an ICLASS,
causing UnboundMethod#== to return false when comparing against the
same method obtained via Module#instance_method.
Resolve ICLASS defined_class in method_eq.
[Bug #21873]
ruby2_keywords warnings: Quote non-UTF8 method names fully
It used to quote only part of the method name because NUL byte in
the method terminates the C string:
```
(irb)> "abcdef".encode("UTF-16LE").bytes
=> [97, 0, 98, 0, 99, 0, 100, 0, 101, 0, 102, 0]
```
```
expected: /abcdef/
actual: warning: Skipping set of ruby2_keywords flag for a (method not defined in Ruby)\n".
```
Do not respect ruby2_keywords on method/proc with post arguments
Previously, ruby2_keywords could be used on a method or proc with
post arguments, but I don't think the behavior is desired:
```ruby
def a(*c, **kw) [c, kw] end
def b(*a, b) a(*a, b) end
ruby2_keywords(:b)
b({foo: 1}, bar: 1)
```
This changes ruby2_keywords to emit a warning and not set the
flag on a method/proc with post arguments.
While here, fix the ruby2_keywords specs for warnings, since they
weren't testing what they should be testing. They all warned
because the method didn't accept a rest argument, not because it
accepted a keyword or keyword rest argument.
* Merge RubyGems-3.5.12 and Bundler-2.5.12
* Merge RubyGems-3.5.13 and Bundler-2.5.13
* Merge RubyGems-3.5.14 and Bundler-2.5.14
* Merge RubyGems-3.5.15 and Bundler-2.5.15
* Merge RubyGems-3.5.16 and Bundler-2.5.16
https://rubyci.s3.amazonaws.com/rhel_zlinux/ruby-3.3/log/20240528T214850Z.fail.html.gz
```
1)
Execution variable $: default $LOAD_PATH entries until sitelibdir included have @gem_prelude_index set FAILED
Expected ["/home/chkbuild/build/20240528T214850Z/mspec/lib/mspec/lib",
"/home/chkbuild/build/20240528T214850Z/mspec/lib",
"./ruby/tool/lib",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/site_ruby/3.3.0",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/site_ruby/3.3.0/s390x-linux",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/site_ruby",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/vendor_ruby/3.3.0",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/vendor_ruby/3.3.0/s390x-linux",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/vendor_ruby",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/3.3.0",
"/home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/3.3.0/s390x-linux"].include? "/home/chkbuild/build/20240528T214850Z/lib/ruby/site_ruby/3.3.0"
to be truthy but was false
/home/chkbuild/build/20240528T214850Z/rubyspec/language/predefined_spec.rb:885:in `block (2 levels) in <top (required)>'
/home/chkbuild/build/20240528T214850Z/rubyspec/language/predefined_spec.rb:846:in `<top (required)>'
```
It does have /home/linux1/chkbuild/tmp/build/20240528T214850Z/lib/ruby/site_ruby/3.3.0,
so it seems actually fine. It seems to be failing due to its setup
issues. Skipping this until we figure out how to fix it.
* Merge RubyGems-3.5.6 and Bundler-2.5.6
* Merge RubyGems-3.5.7 and Bundler-2.5.7
* Merge RubyGems-3.5.8 and Bundler-2.5.8
* Partly reverted about https://github.com/rubygems/rubygems/pull/7483
* Merge RubyGems-3.5.9 and Bundler-2.5.9
rb_obj_setup: do not copy RUBY_FL_SEEN_OBJ_ID
[Bug #20250]
We're seting up a new instance, so it never had an associated
object_id.
proc.c: get rid of `CLONESETUP`
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
[Bug #20253]
All the way down to Ruby 1.9, `Proc`, `Method`, `UnboundMethod`
and `Binding` always had their own specific clone and dup routine.
This caused various discrepancies with how other objects behave
on `dup` and `clone. [Bug #20250], [Bug #20253].
This commit get rid of `CLONESETUP` and use the the same codepath
as all other types, so ensure consistency.
NB: It's still not accepting the `freeze` keyword argument on `clone`.
Co-Authored-By: Étienne Barrié <etienne.barrie@gmail.com>
* Merge RubyGems-3.5.4 and Bundler-2.5.4
* Merge RubyGems-3.5.5 and Bundler-2.5.5
* Make tests play with upstream Ruby tests
CI broke in https://github.com/ruby/ruby/pull/9604 because if any Ruby
tests run `require 'net/http'`, they will pollute the
`$LOADED_FEATURES` for the RubyGems tests. We can fix this by renaming
the test default gem from `net-http` to `my-http`.
See https://github.com/rubygems/rubygems/pull/7379#issuecomment-1901241299
for more details.
---------
Co-authored-by: Stan Hu <stanhu@gmail.com>
A default Bundler copy with vendored net-http is causing a Bundler spec
to end up loading that copy instead of the copy of Bundler under test.
This is because of vcr, but we don't really need vcr in the command
that's failing, so I avoided loading it.
https://github.com/rubygems/rubygems/commit/304ab6e1b9
