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
If a gem is specified in the Gemfile (or resolved as a transitive
dependency), it's always resolved from remote/installed sources. Default
gems are only used as a fallback for gems not included in the bundle.
I believe this leads to more consistent behavior and more portable apps,
since all gems will be installed to the configured bundle path,
regardless of whether they are default gems or not.
https://github.com/rubygems/rubygems/commit/091b4fcf2b
This makes bundler consistent with all other gems, and makes the default
installation of Bundler in the release package look like any other
bundler installation.
Before (on preview3, for example), Bundler executable is installed at:
lib/ruby/gems/3.3.0+0/gems/bundler-2.5.0.dev/libexec/bundle
Now it's installed in the standard location:
lib/ruby/gems/3.3.0+0/gems/bundler-2.5.0.dev/exe/bundle
So generate_index can be implemented with dependencies, such as the compact index
Took this approach from feedback in https://github.com/rubygems/rubygems/pull/6853
Running `gem generate_index` by default will use an installed rubygems-generate_index, or install and then use the command from the gem
Apply suggestions from code review
https://github.com/rubygems/rubygems/commit/fc1cb9bc9e
Co-authored-by: Hiroshi SHIBATA <hsbt@ruby-lang.org>
Since Ruby 3.0, Ruby has passed a keyword splat as a regular
argument in the case of a call to a Ruby method where the
method does not accept keyword arguments, if the method
call does not contain an argument splat:
```ruby
def self.f(obj) obj end
def self.fs(*obj) obj[0] end
h = {a: 1}
f(**h).equal?(h) # Before: true; After: false
fs(**h).equal?(h) # Before: true; After: false
a = []
f(*a, **h).equal?(h) # Before and After: false
fs(*a, **h).equal?(h) # Before and After: false
```
The fact that the behavior differs when passing an empty
argument splat makes it obvious that something is not
working the way it is intended. Ruby 2 always copied
the keyword splat hash, and that is the expected behavior
in Ruby 3.
This bug is because of a missed check in setup_parameters_complex.
If the keyword splat passed is not mutable, then it points to
an existing object and not a new object, and therefore it must
be copied.
Now, there are 3 specs for the broken behavior of directly
using the keyword splatted hash. Fix two specs and add a
new version guard. Do not keep the specs for the broken
behavior for earlier Ruby versions, in case this fix is
backported. For the ruby2_keywords spec, just remove the
related line, since that line is unrelated to what the
spec is testing.
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
Additionally, the result is memoized, as it's used twice in a row.
This change does result in a net behavioral diff, as the list of ENVs
being checked has been updated (now includes buildkite, taskcluster,
cirrus, dsari, and drops buildbox and snap)
https://github.com/rubygems/rubygems/commit/3fb445a5a1
Because bundler needs to support older versions of rubygems, we can't
actually rely on Gem::CIDetector (yet - in a year or so they might be
able to consolidate, if they don't change futher). So we're copying it
into the Bundler:: namespace, and enforcing that they stay completely in
sync with a test. No other tests are needed, since Gem::CIDetector is
already tested, and this is and will remain identical.
https://github.com/rubygems/rubygems/commit/abc67f0da1
Looks for the CHECKSUMS section in the lockfile, activating the feature
only if the section exists. Without a CHECKSUMS section, Bundler will
continue as normal, validating checksums when gems are installed while
checksums from the compact index are present.
https://github.com/rubygems/rubygems/commit/2353cc93a4
The original ripper test was very stable as the output didn't change. Prism is under active development and changing their output shouldn't cause a failure to the ruby/ruby test suite like https://github.com/ruby/ruby/actions/runs/7104601478/job/19339940315.
This commit moves from checking exact output to asserting that the string we get back is not empty. This should give the same level of confidence that some error message was caught, and is less brittle.
https://github.com/ruby/syntax_suggest/commit/4b6abb763e
Prism will be the parser in Ruby 3.3. We need to support 3.0+ so we will have to "dual boot" both parsers.
Todo:
- LexAll to support Prism lex output
- Add tests that exercise both Ripper and prism codepaths on CI
- Handle https://github.com/ruby/prism/issues/1972 in `ripper_errors.rb`
- Update docs to not mention Ripper explicitly
- Consider different/cleaner APIs for separating out Ripper and Prism
https://github.com/ruby/syntax_suggest/commit/a7d6991cc4
Currently, auto-install with git gems fails, when
it would succeed with a rubygems-source gem
Fix the issue by doing the same fallback for git errors as we do for
missing gems, the git errors should only bubble up in these cases when
the gem is not installed, meaning we want to go through the install flow
(and any persistent errors will be raised through there)
https://github.com/rubygems/rubygems/commit/e25a339f7a
This changes the CompactIndexClient to store etags received from the
compact index in separate files rather than relying on the MD5 checksum
of the file as the etag.
Smoothes the upgrade from md5 etags to opaque by generating them when no
etag file exists. This should reduce the initial impact of changing the
caching behavior by reducing cache misses when the MD5 etag is the same.
Eventually, the MD5 behavior should be retired and the etag should be
considered completely opaque with no assumption that MD5 would match.
Hiroshi SHIBATA
David Rodríguez
Martin Emde
Samuel Giddins
Franz Liedke
Yuji Yaginuma
Jeremy Evans
Eric Mueller
Schneems
Misaki Shioi
Tema Bolshakov
Benoit Daloze
Josef Šimánek