| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |\
| |
| |
| | |
Version 1.15.0.pre.2
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Ensure pre-existing Git caches are updated from remote sources
This will be my first contribution to Bundler. Thanks for the amazing dev setup guides!
This fixes issue #5423 where Git sources would not be updated if they were already cached in `vendor/cache`.
When we build a Git source from Gemfile.lock, we store the revision listed in the lockfile. However, when unlocking the source for an update, we don't clear that revision, and the update proceeds to grab the stale version from `vendor/cache`. Interestingly, this behaviour only occurs when updating a specific gem, i.e. `bundle update timecop`. This bug isn't present when doing a global `bundle update`, possibly because of the differences in how we build up the Definition between the two methods.
Closes #5423.
(cherry picked from commit 334b901de575ca4b6b88e164106e2a52f4f4bff9)
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Avoid duplicating specs in the lockfile after updating w/ the gem uninstalled
Closes https://github.com/bundler/bundler/issues/5599
I decided to not compare using `full_name` in `SpecSet#merge` for the sake of performance, since `#full_name` is uncached
(cherry picked from commit 8e2cd4ac9b1b0c3459b5969af364b318821121e5)
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Apply source and git_version to stub spec extension
Fixes https://github.com/bundler/bundler/issues/5594
`stub_specification` doesn't know what `full_name` should be because we don't have a record of the source, so it resolves to `cityhash-0.6.0` rather than `cityhash-3cfc7d01f333`
This makes `StubSpecification` aware of the source to we can still avoid loading the full spec when needed. This is done by bringing remote_specs `attr_accessor :source` to `stub_spec`, and assigning source to the `stub`.
I am probably missing some edge cases to check for though 🙈 Need to write specs.
cc @segiddins
(cherry picked from commit dcddaf98a7201483277b71ae1fb7904a15a0cc9f)
|
| | | |
|
| |\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Ensure pre-existing Git caches are updated from remote sources
This will be my first contribution to Bundler. Thanks for the amazing dev setup guides!
This fixes issue #5423 where Git sources would not be updated if they were already cached in `vendor/cache`.
When we build a Git source from Gemfile.lock, we store the revision listed in the lockfile. However, when unlocking the source for an update, we don't clear that revision, and the update proceeds to grab the stale version from `vendor/cache`. Interestingly, this behaviour only occurs when updating a specific gem, i.e. `bundle update timecop`. This bug isn't present when doing a global `bundle update`, possibly because of the differences in how we build up the Definition between the two methods.
Closes #5423.
|
| | | | |
|
| | | | |
|
| |\ \ \
| | | |
| | | |
| | | |
| | | |
| | | | |
Dsl improvements
This PR contains a couple of small improvements to help remove code complexity in the DSL. Refer to the commit messages for context.
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| |\ \ \ \
| |_|/ /
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Avoid duplicating specs in the lockfile after updating w/ the gem uninstalled
Closes https://github.com/bundler/bundler/issues/5599
I decided to not compare using `full_name` in `SpecSet#merge` for the sake of performance, since `#full_name` is uncached
|
| | | | | |
|
| |\ \ \ \
| | | | |
| | | | |
| | | | | |
Fix exec_spec on ruby-head
|
| | | | | | |
|
| | |/ / / |
|
| |\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Apply source and git_version to stub spec extension
Fixes https://github.com/bundler/bundler/issues/5594
`stub_specification` doesn't know what `full_name` should be because we don't have a record of the source, so it resolves to `cityhash-0.6.0` rather than `cityhash-3cfc7d01f333`
This makes `StubSpecification` aware of the source to we can still avoid loading the full spec when needed. This is done by bringing remote_specs `attr_accessor :source` to `stub_spec`, and assigning source to the `stub`.
I am probably missing some edge cases to check for though 🙈 Need to write specs.
cc @segiddins
|
| | | | | | |
|
| | | | | | |
|
| | |/ / / |
|
| |\ \ \ \
| |/ / /
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
move implementation of Bundler::Env to class methods
This is to a comment i made in #5589
Currently to print the Bundler env report you need to need to invoke it
via:
Bundler::Env.new.report
This change removes the need to instantiate an instance and just call
a class method of the same name:
Bundler::Env.report
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently to print the Bundler env report you need to need to invoke it
via:
Bundler::Env.new.report
This change removes the need to instantiate an instance and just call
the class method of the same name:
Bundler::Env.report
|
| |\ \ \ \
| | |_|/
| |/| |
| | | | |
Version 1.15.0.pre.1
|
| |/ / / |
|
| |\ \ \
| |/ /
| | |
| | |
| | |
| | | |
[Env] Default to always printing gemfile & gemspecs
This will save a lot of time spent asking for them, and is something I thought we'd already done
|
| |/ / |
|
| |\ \
| |/
|/|
| |
| |
| |
| |
| | |
Speed up spec sets
Nothing major, but why not ¯\_(ツ)_/¯
Closes #5057
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| |\ \
| |/
|/|
| |
| |
| |
| |
| |
| |
| |
| | |
Bundler config parseable flag
This PR is continuing on from #4919. This adds an option to `bundle config` called `parseable` that prints the value of config without the verbose. Example:
$ bundle config foo --parseable
bar
Closes #4919
|
| | |
| |
| |
| | |
existing config
|
| | |
| |
| |
| | |
value
|
| | | |
|
| | | |
|
| | | |
|
| | |\
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
bundler-config-parseable-flag
* upstream/master: (522 commits)
add tests
fix EOL
Require socket in the file that needs it
use Gem::Version.correct? instead of catching exceptions
remove rails 2.3 from ISSUES guide
quote empty PATH
turns out linux needs env vars to be quoted (even when empty)
run specs without git in the env
add specs to test gem command when git is not installed (note: this won't actually work at the moment)
dont create .gitignore as well if git is not installed in the gem command
dont init new git repo unless git is installed on gem command
fix spacing between paragraphs in generated READMEs for new gems
Account for differing behavior in hash#select between 1.8.7 and 1.9.3+
.map -> .each, as we are not using return value
Only uninstall plugins that we attempted to install.
Documentation at the correct place
Double quotes and small documentation
--key parameter can be set via bundle config gem.push_key
Add a spec for nested bundle exec with a gem that shadows a default gem
Allow installing a gem from an arbitrary repo in the specs
...
|
| | | | |
|
| | | | |
|
| | | | |
|
