This change prevents default gems from inadvertently depending on
bundled gems. This issue was discovered by being able to
`require "rbs"` from test/irb.
During the build, Ruby has special logic to serialize its own builtin
module to disk using the binary iseq format during the build (I assume
for speed so it doesn't have to parse builtin every time it starts
up).
However, since iseq format is architecture-specific, when building on
x86_64 for universal x86_64 + arm64, the serialized builtin module is
written with the x86_64 architecture of the build machine, which fails
this check whenever ruby imports the builtin module on arm64:
1fdaa06660/compile.c (L13243)
Thankfully, there's logic to disable this feature for cross-compiled builds:
1fdaa06660/builtin.c (L6)
This disables the iseq logic for universal builds as well.
Fixes [Bug #18286]
The `test_build_extensions` make assumptions about return value of
`Gem.install_extension_in_lib`. Givent that RubyGems allow to override
this method via `defaults.rb` / `operating_system.rb`, this test might
not always pass. This change makes sure that the
`Gem.install_extension_in_lib` is predictable.
https://github.com/rubygems/rubygems/commit/e83d0a5da8
Daily Bundler CI against ruby-head is failing because ruby-head now
uses bigdecimal 3.1.5, so that gets locked by this spec.
This change should make the test stable until bigdecimal 99.1.5 is
bundled with Ruby :)
https://github.com/rubygems/rubygems/commit/830326041f
We know where it is in the iseq local table, because we forced it using
`iseq_set_exception_local_table` when compiling the BEGIN_NODE, so we
can always look it up from the same place.
Co-Authored-By Peter Zhu <peter@peterzhu.ca>
Instead of storing the delegate in @tmpfile, use __getobj__, since
delegate library already handles dup/clone for that. Copy the
unlinked, mode, and opts instance variables to the returned object
when using dup/clone.
Split the close/unlink finalizer into two finalizers. The close
finalizer always closes when any Tempfile instance is GCed, since
each Tempfile instance uses a separate file descriptor. The unlink
finalizer unlinks only when the original and all duped/cloned
Tempfiles are GCed, since all share the same path.
For Tempfile#open, undefine the close finalizer after closing the
current file, the redefine the close finalizer with the new file.
Fixes [Bug #19441]
https://github.com/ruby/tempfile/commit/dafabf9c7b
before this change, creating a regex from a regex with a named capture, Regexp.new(/(?<name>)/), causes memory to be shared between the two named capture groups which can cause a segfault if the original is GCed.
Since #6945 the extension dir changed to Gem::BasicSpecification's implementation, we didn't hook that in rubygems_ext.rb. So for universal rubies, we ended up using the universal platform name when installing, but arch replaced platform name when checking. This lead to native extensions can never be correctly installed on universal rubies.
Hook Gem::BasicSpecifications so the behavior is consistent on installing and checking.
https://github.com/rubygems/rubygems/commit/8d699ed096
Right now the `rb_shape_get_next` shape caller need to
first check if there is capacity left, and if not call
`rb_shape_transition_shape_capa` before it can call `rb_shape_get_next`.
And on each of these it needs to checks if we got a TOO_COMPLEX
back.
All this logic is duplicated in the interpreter, YJIT and RJIT.
Instead we can have `rb_shape_get_next` do the capacity transition
when needed. The caller can compare the old and new shapes capacity
to know if resizing is needed. It also can check for TOO_COMPLEX
only once.
(https://github.com/ruby/irb/pull/708)
* Add completor using prism and rbs
* Add TypeCompletion test
* Switchable completors: RegexpCompletor and TypeCompletion::Completor
* Add completion info to irb_info
* Complete reserved words
* Fix [*] (*) {**} and prism's change of KeywordParameterNode
* Fix require, frozen_string_literal
* Drop prism<=0.16.0 support
* Add Completor.last_completion_error for debug report
* Retrieve `self` and `Module.nesting` in more safe way
* Support BasicObject
* Handle lvar and ivar get exception correctly
* Skip ivar reference test of non-self object in ruby < 3.2
* BaseScope to RootScope, move method objects constant under Methods
* Remove unused Splat struct
* Drop deeply nested array/hash type calculation from actual object. Now, calculation depth is 1
* Refactor loading rbs in test, change preload_in_thread not to cache Thread object
* Use new option added in prism 0.17.1 to parse code with localvars
* Add Prism version check and warn when :type completor cannot be enabled
* build_type_completor should skip truffleruby (because endless method definition is not supported)
https://github.com/ruby/irb/commit/1048c7ed7a
Comparing file paths as strings may not work well for some reasons,
symlink, relative `__FILE__`, etc.
Some alternatives are possible: comparing with `File.realpath`, or
with `File.identical?`, it should be most robust to escape the target
string contained within this file itself.
Nobuyoshi Nakada
Yusuke Endoh
Ben Hamilton
Vít Ondruch
David Rodríguez
Matt Valentine-House
Kevin Newton
Jemma Issroff
Peter Zhu
Jeremy Evans
Mau Magnaguagno
Takashi Kokubun
git
Burdette Lamar
Adam Hess
Adam Hess
Han Young
Jean Boussier
Sutou Kouhei
tomoya ishida
dependabot[bot]
Jun Aruga